The Pennsylvania State University. The Graduate School. College of Medicine

Transcription

1 The Pennsylvania State University The Graduate School College of Medicine THE ROLE OF TIGHTLY COORDINATED NEURONAL FIRING IN SOMATOSENSORY PROCESSING A Thesis in Neuroscience By Stephane A. Roy Copyright 2001 Stephane A. Roy Submitted in Partial Fulfillment Of the Requirements For the Degree of Doctor of Philosophy May 2001

2 We approve the thesis of Stephane A. Roy. Date of Signature Kevin D. Alloway Associate Professor of Neuroscience & Anatomy Thesis Advisor Chair of Committee Steven P. Dear Assistant Professor of Neuroscience & Anatomy Assistant Professor of Acoustics Thomas C. Pritchard Assistant Professor of Behavioral Science Seth Wolpert Associate Professor of Engineering Robert J. Milner Professor of Neuroscience & Anatomy Chair of the Graduate Program in Neuroscience

3 iii ABSTRACT The fundamental observation that single neurons encode information in their average rate of firing has shaped much of our understanding of nervous system function. However, it has become apparent that in addition to firing rates, the precise timing of individual spikes on a millisecond scale may play a critical role in how information is processed among populations of neurons. In particular, the near-simultaneous firing of neurons within a cortical area or across cortical areas has received much attention as a potential sensory code. These experiments investigated the potential role of synchronous firing in the somatosensory system. Three main questions were addressed: whether or not synchronization was present among local populations of neurons in primary somatosensory cortex during sensory stimulation; if synchronization could serve as a signal to bind the simultaneous stimulus responses in primary and secondary somatosensory cortex; and can cortical neurons distinguish between synchronous and asynchronous presynaptic inputs from the thalamus. The results presented here show that within SI, local populations of neurons become synchronized during sensory stimulation to varying degrees depending on the nature of the stimulus. In addition, neurons in SI and SII with overlapping receptive fields become tightly synchronized when responding to a single stimulus. Finally, we have also addressed the next important step in establishing the importance of synchronized firing in sensory processing: demonstrating that cortical neurons possess a mechanism to decode the information in synchronous spikes.

4 iv TABLE OF CONTENTS LIST OF FIGURES... viii LIST OF TABLES...x ACKNOWLEDGEMENTS...xi Chapter 1. Literature Review Coordinated Firing in Sensory Processing The theoretical framework Synchronized firing and the formation of neural assemblies Synchrony as a sensory code Mechanisms of synchronization Are Synchronous spikes special? Synchronous firing in Somatosensory Cortex Parallel processing of sensory information in SI and SII Summary of rationale...8 Chapter 2. Synchronization of Local Neural Networks in the Somatosensory Cortex: A Comparison of Stationary and Moving Stimuli Introduction Materials and Methods Cortical electrophysiology Cutaneous stimulation Analysis of neuronal responses Autocorrelation analysis Histology...18

5 v 2.3. Results Modulation of SI synchronization by stationary air jets Modulation of SI synchronization by moving air jets Comparison of synchronization induced by stationary and moving air jets Effects of electrode separation on stimulus-induced cortical synchronization Lack of oscillations during stimulus-induced synchronization Discussion Anatomic factors affecting synchrony in SI cortex Synchronization in SI cortex and sensory coding Parallels with other sensory systems Comparisons of multiple and single neuron responses Interpretation of raw and shift-corrected CCGs...47 Chapter 3. Long-Range Cortical Synchronization Without Concomitant Oscillations in the Somatosensory System Introduction Methods Surgery Electrophysiology Cutaneous Stimulation Cross-correlation analysis Analysis of Neuronal Oscillations Results...61

6 vi Synchronization of Single Unit Activity in SI and SII Synchronization of Multiunit Activity in SI and SII Distribution of Single and Multiple Neuron Synchronization Incidence of Neuronal Oscillations in the ACGs Incidence of Oscillations in the CCGs Discussion Variable and Transient Incidence of Synchronized Oscillations Periodic and Aperiodic Cortical Synchronization Plausible Mechanisms of SI-SII Synchronization Hypothetical Roles for Synchronization in SI and SII...83 Chapter 4. Coincidence detection or temporal integration? What the neurons in somatosensory cortex are doing Introduction Materials and Methods Animal Preparation Electrophysiology Cutaneous stimulation Cross-correlation analysis Results Snowflake Analysis Conditional Cross-correlation Analysis Interspike Interval Analysis Discussion...101

7 vii Chapter 5. Conclusions & Future Directions Summary of Conclusions Synchronization within SI, revisited The role of synchronized firing across cortical areas Synchronous spiking in networks Important considerations Future Directions Bibliography...112

8 viii LIST OF FIGURES Figure 2.1 Responses to stationary air jets by a pair of primary (SI) somatosensory cortex neurons that had overlapping receptive fields (RFs) and were separated by a distance of 300 µm Figure 2.2 Scatter plots comparing the parameters of cortical synchronization for 87 neuron pairs that had correlated discharges during spontaneous activity and stationary airjet stimulation Figure 2.3 Responses of the same pair of SI neurons shown in Figure 2.1 but during subsequent stimulation with a moving air jet...24 Figure 2.4 Scatter plots comparing the parameters of cortical synchronization for 88 neuron pairs that had correlated discharges during spontaneous activity and in response to moving air-jet stimulation...25 Figure 2.5. Comparison of moving and stationary air jets on neuronal synchronization in SI cortex Figure 2.6 Comparison of moving and stationary air jets on the synchronization rate of the same multiunit responses shown in Figure Figure 2.7 Comparison of stationary and moving airjets on synchronization rate Figure 2.8 Comparison of the proportion of activity that was correlated during stationary and moving air jets...32 Figure 2.9. Comparison of stationary and moving air jets on the timing of synchronized activity Figure 2.10 Changes in the rate of synchronization produced by stationary and moving air jets as a function of distance between electrodes Figure 2.11 Changes in the proportion of synchronized activity produced by stationary and moving air jets as a function of the distance between electrodes...36 Figure 2.12 Changes in the timing of synchronized activity produced by stationary and moving air jets as a function of distance between electrodes...37 Figure 3.1 Method for detecting neuronal oscillations...58 Figure 3.2 Synchronized activity in a pair of SI and SII neurons in which one neuron contained weak oscillations in the gamma frequency range....62

9 ix Figure 3.3 Synchronized activity in a pair of SI and SII neurons (A-41) that did not oscillate in the gamma frequency range...65 Figure 3.4 Synchronized multiunit responses in SI and SII (A-130) without concomitant oscillations Figure 3.5 Cumulative distributions illustrating the strength of spontaneous and stimulusinduced synchronization across SI and SII Figure 3.6 Precision of temporal synchronization for single unit and multiunit stimulusinduced responses in SI and SII...71 Figure 3.7 Distribution of neuronal oscillations according to frequency Figure 3.8 Strength of neuronal oscillations...72 Figure 3.9 Strength of synchronized activity in SI and SII as a function of the presence or absence of oscillations in the constituent neuronal responses...74 Figure 3.10 Trial-by-trial analysis of oscillatory responses in experiment A Figure 3.11 Neural CCGs illustrating the degree of synchronization in experiment A130 for those trials classified as containing no oscillations (n = 59), oscillations in SI only (n = 8), in SII only (n = 29), or in both cortical areas (n = 4) Figure 4.1 Representative experiment (TC12) illustrating relationship between thalamic synchronization and thalamocortical coordination Figure 4.2 Cortical responses to varying amounts of thalamic synchronization Figure 4.3. Conditional cross-correlation analysis of synchronous and asynchronous thalamic discharges on neuronal responses in SII cortex Figure 4.4 Changes in thalamocortical efficacy as a function of search interval duration Figure 4.5 Thalamocortical efficacy as function of interneuronal interspike intervals Figure 4.6 Thalamocortical efficacy as a function of interneuronal ISIs....99

10 x LIST OF TABLES Table 2.1 Spontaneous and stimulus-induced neuronal firing rates in SI cortex...19 Table 2.2 Probability of synchronization across SI recording sites separated by varying distances...34 Table 3.1 Incidence of oscillations during SI-SII synchronization...73 Table 3.2. Stimulus-induced activity of synchronized neurons in SI and SII...74

11 xi ACKNOWLEDGEMENTS I owe my advisor, Kevin Alloway, a great deal of thanks for his continual insight and guidance. He taught me the ability to recognize the most important points from large and potentially obscure pools of data, which was invaluable in determining the direction of experiments. I hope that this clarity of interpretation and concentration of effort are skills that I can continue to use in the future. Numerous classmates and friends have provided invaluable company and entertainment outside the lab: Phil, Joe, Nicole, Josh, Zak, Mike, and many others throughout the years; Ed, John, Bruce and all the others who were always ready to play some ball; and the Medwings, to name a few. I hope I have had as positive an impact on their experience as they did on mine. My family has always been unconditionally supportive which I know has been more important to my success than I have been able to fully appreciate. Throughout the ups and downs of the last few years I have been privileged to have Brenna by my side; she has made this entire effort so much more enjoyable. I look forward to her continued companionship in all the years to come.

12 xii The really efficient laborer will be found not to crowd his day with work, but will saunter to his task surrounded by a wide halo of ease and leisure. Henry David Thoreau

13 1 Chapter 1. Literature Review 1.1 Coordinated Firing in Sensory Processing The fundamental observation that single neurons encode information in their average firing rate has shaped much of our understanding of nervous system function (for review, see decharms and Zador 2000). It has become apparent, however, that in addition to firing rates, the precise timing of individual spikes on a millisecond scale may play a critical role in how information is processed by populations of neurons. In particular, the near-simultaneous firing of neurons within a cortical area or across cortical areas has received much attention as a potential sensory code. Experimental results highlighting the significance of synchronous spiking fall largely into two closely related categories. The first set consists of experiments showing that the incidence of synchronous discharges between neurons can code sensory information in parallel with, and in some cases independent of, firing rate. The second set of experiments revolve around the idea that synchronous firing can serve as a label for grouping distributed neuronal responses that are involved in closely related tasks. Both schemes require that the nervous system possess some mechanism for reading synchronous spikes The theoretical framework Within a cortical area the response to a single perceptual object frequently spans many neurons with non-overlapping receptive fields (Gray et al. 1989; Engel et al. 1990). In addition, sensory pathways typically diverge so that by the time a peripheral stimulus

14 2 is conveyed to cortex it can elicit responses among neuronal populations with different response properties in several areas of cortex. An important question in sensory physiology has been how the brain is able to consolidate these distributed responses into a single percept (for an overview of the problem see Treisman 1996). One theory that has received much support in recent decades is the role of synchronized firing as a mechanism for grouping neuronal populations. The potential significance of synchronous spikes in population binding was first put forth almost 20 years ago when Cristoph von der Malsburg (von der Malsburg 1981) suggested that neurons could form assemblies by engaging in synchronous spiking activity. According to this theory, groups of neurons could form multiple assemblies to represent different contiguous stimulus features; in the visual system, for example, separate assemblies might exist for foreground & background objects. Membership in an assembly could be easily distinguished by the precise firing times of neurons relative to each other, and cells could form dynamic assemblies depending on attributes of the stimulus. This theory provided a potentially important mechanism in the representation of the myriad of possible stimuli available in the natural environment. The idea that synchronously active neurons form functional assemblies was extended to suggest that if layers of neurons in a network were connected in a feedforward manner, then a volley of synchronous spikes in the first layer or assembly would be transmitted to the second layer, and so on, forming a synfire chain (Abeles 1982). An important condition that exists with this idea is that neurons respond preferentially to simultaneous presynaptic inputs. This was believed to be the case based on mathematical

15 3 models of cortical networks (Abeles 1982). These theories, while difficult to test at the time, predicted that synchronization plays an important part not only in feature binding and object recognition during sensory processing, but also more generally in the transmission of information throughout cortical areas. Subsequently, many experiments have investigated the occurrence of synchronized firing in the nervous system and its dependence on a variety of factors including stimulus configuration and behavioral state Synchronized firing and the formation of neural assemblies The suggestion that synchronized firing could serve to tag assemblies of neurons on a dynamic basis was eventually reinforced by observations in the visual system that neurons in striate cortex with non-overlapping receptive fields and similar orientation preferences could display synchronous oscillations when presented with a single collinear stimulus, and that this synchronization disappeared when two separate stimuli were present (Gray et al. 1989). Subsequent experiments in the visual system showed that synchronization occurred among neurons in different cortical areas (Engel et al. 1991) and between hemispheres (Engel et al. 1991). The notion that neurons with overlapping receptive fields could form selective assemblies with each other depending on stimulus attributes was also confirmed experimentally (Engel et al. 1991). These experiments set the foundation for studies in a variety of cortical areas that investigated the role of synchronized firing in central processes Synchrony as a sensory code It has been suggested that the precise timing or occurrence of correlated discharges between neurons may vary substantially independent of changes in the firing

16 4 rates of neurons. Thus precise spike timing in populations of neurons could serve as an additional means to convey information about stimulus properties, as in the example of feature binding above. Several experiments have shown that synchronized spikes code certain stimulus parameters better than firing rate alone (Ahissar et al. 1992; decharms and Merzenich 1996; Dan et al. 1998). In addition, synchronization can be more closely correlated with attention (Steinmetz et al. 2000) or behavioral states (Vaadia et al. 1995; Riehle et al. 1997) than firing rates. One confounding factor in the effort to show that synchronized firing carries additional information is that an increase in the number of correlated discharges is usually accompanied by an increase in the firing rates of the neurons. This has prompted the criticism that synchronous firing is an unavoidable side effect of high firing rates and does not contain any additional information about the sensory environment or the organism. Consequently, many investigators have gone to great lengths when quantifying the amount synchronized firing to compensate for the expected changes in synchronization that would be due to either stimulus locking or random chance, and analyze only correlations of neural origin. In some cases experiments have now demonstrated variations in synchronization that correlate with behavioral states or stimulus features in the absence of measurable changes in firing rate (Vaadia et al. 1995; decharms and Merzenich 1996; Riehle et al. 1997). These observations provide the strongest evidence that precise spike timing can indeed contain information that is not available by merely looking at firing rates.

17 5 It is important to note that methods to remove stimulus-locked firings among pairs of neurons cannot distinguish between expected events and coincident events that might code additional information about the stimulus. For example, the commonly used shiftpredictor (Perkel et al. 1967) assumes that the synchronous activity between a pair of neurons is a linear sum of stimulus-locked events, random coincidences, and any additional correlation of neural origin. Subtraction of the shift predictor corrects for the number of synchronous events, but does not distinguish between the potential mechanisms generating each pair of synchronized spikes. If sensory information is encoded in only a portion of the synchronized events, then this subtraction probably makes it more difficult to ascertain the role of synchronous firing by removing important occurrences. Furthermore, it is unclear that the nervous system can make this distinction between expected & unexpected synchrony. As a result it is becoming more common to analyze raw correlations in addition to neural or corrected correlations, because these represent all the events that are available to the nervous system during sensory processing (Eggermont and Mossop 1998; Roy and Alloway 1999). While this technique may include randomly occurring coincident events, if the aim of the study is to investigate the impact of coordinated firing then all coincident events should be included. 1.2 Mechanisms of synchronization The precisely timed action potentials required for synchronized firing can arise from several mechanisms depending on the underlying anatomical connectivity of the network. If two neurons receive excitatory inputs from the same presynaptic neuron or region, they should tend to fire at the same time, because they receive a similar pattern of

18 6 EPSPs. While numerous modeling studies have confirmed that common input can produce synchronization, it is difficult to demonstrate directly that this is a mechanism for generating synchronized firing in vivo. Most experiments conclude that synchronized firing is the result of common input based on the underlying anatomical connectivity and the characteristic tight timing of common-input induced synchronization. In primary cortical areas, experimental observations have concluded that synchronization due to common input could originate from thalamic or other cortical regions (Munk et al. 1995). Monosynaptic connections between a pair of neurons can also yield precisely correlated near-synchronous discharges reflecting the conduction and synaptic delays between the neurons. It is thought, however, that these horizontal connections can also help to cause synchronization when the underlying units oscillate independently (Ts'o et al. 1986; Eckhorn et al. 1988). These synchronous oscillations have received a large amount of support in the visual system as the dominant mechanism for generating synchrony. Under certain stimulus configurations neurons oscillate in response to a stimulus (Eckhorn et al. 1988), and neurons with non-overlapping receptive fields but similar orientation preferences that are presented with a single collinear stimulus will oscillate in phase with one another (Gray and Singer 1989). This is presumably mediated via horizontal connections that selectively phase lock the oscillations of each neuron in the gamma frequency range (20-80 Hz). Other experiments have shown, however, that synchronous firing under very similar stimulus configurations can occur without oscillations (Nowak et al. 1995).

19 7 1.3 Are Synchronous spikes special? If synchronous neuronal firing is a code that carries information in sensory systems or tags groups of neuronal assemblies, then there must be a mechanism by which neurons decode or interpret this information. The most obvious mechanism for the brain to distinguish between synchronous and asynchronous spikes is the integrative properties of neurons receiving convergent inputs from cells that may or may not be synchronized. The ability of a pyramidal cortical neuron to distinguish between closely correlated inputs in vivo has been the subject of extensive debate, however (Abeles 1982; Shadlen and Newsome 1994). The primary issues can be divided into two questions. First, is there an ideal time interval for near-coincident inputs (or EPSPs) to evoke a post-synaptic spike? If synchronous inputs have greater impact on postsynaptic responsiveness than asynchronous inputs, then what is the time interval in which this effect is present? If this interval is less than the interspike interval of the postsynaptic neuron in question, then synchronous spikes will have different impacts on the firing time of the postsynaptic neuron than asynchronous spikes (Konig et al. 1996). Second, do groups of neurons possess the ability to show significant synchronization on this time scale? While these issues do not address how distributed synchronous populations are related to perception, they have important implications for the mechanisms underlying the transmission of sensory information. 1.4 Synchronous firing in Somatosensory Cortex Until recently only one study had examined the presence of synchronous firing in somatosensory cortex (Metherate and Dykes 1985). These experiments showed that

20 8 cortical neurons with similar response properties exhibited synchronous firing while neurons with different response properties did not. This study did not compare the responses of the neurons under different stimulus conditions. Thus while it had been demonstrated that neurons in somatosensory cortex could become synchronized with each other, it was still unclear to what extent the synchronization might code additional information about the stimulus Parallel processing of sensory information in SI and SII In the feline somatosensory system, both SI and the second somatosensory area (SII) receive significant somatotopically organized projections from the ventrobasal complex of the thalamus (Landry & Deschênes 1981;Hand & Morrison 1970;Niimi et al. 1987;Hand & Morrison 1970). These projections come from neurons in overlapping regions of VPL, and some thalamocortical neurons send axons to both SI & SII (Yanagihara et al. 1987). Physiological experiments confirm this parallel organization. Inactivation of cortical area SI by undercutting thalamocortical afferents (Manzoni et al. 1979), lidocaine anesthesia (Burton & Robinson 1987) or cooling (Turman et al. 1992) rarely leads to unresponsiveness of SII neurons with cutaneous receptive fields. This parallel organization has been reported for proprioceptive inputs as well (Mackie et al. 1996). Thus, the thalamic input is sufficient to drive SII neurons in the absence of SI association projections. 1.5 Summary of rationale In summary, the experiments included in this thesis were prompted by several fundamental observations. First, in light of the lack of data on the potential role of

21 9 synchronization as a sensory code in somatosensory cortex, we set out to determine if synchronized firing among local populations of neurons in SI varied during stimulus presentation (Chapter 2). Second, if synchrony binds multiple neural representations and is a universal principle common to all sensory areas, then neurons in SI and SII that respond to the same stimulus should show synchronous firing (Chapter 3). Finally, the ability of synchronized firing to increase the responsiveness of postsynaptic neurons was investigated in the thalamocortical system (Chapter 4). General conclusions and future directions are summarized in Chapter 5.

22 10 Chapter 2. Synchronization of Local Neural Networks in the Somatosensory Cortex: A Comparison of Stationary and Moving Stimuli 2.1. Introduction Recent work in the visual system suggests that perceptual objects are represented by synchronous activity among populations of neurons that respond to the individual components of the object (for reviews, see Singer and Gray 1995; Singer et al. 1997). Cross-correlation analysis of neuronal activity in striate cortex, for example, has shown that neurons representing adjacent parts of the visual field become synchronized if they have similar orientation preferences and are activated by a single bar of light that stretches across their receptive fields (Gray et al. 1989). This result has prompted the hypothesis that cortical synchronization represents a dynamic mechanism for increasing the salience of activity among cortical neurons that respond to different segments of the same linear stimulus. According to this view, synchronization in striate cortex is mediated by intracortical connections that provide a substrate for linking separate neural populations into functional assemblies for the perception of contours and other stimulus features that have spatial continuity (Ts'o et al. 1986; Singer and Gray 1995). If neuronal synchronization is a universal principle of cortical physiology that underlies aspects of perception in all sensory modalities, then stimulus-induced synchronization should occur in other sensory regions, including the somatosensory cortex. In support of this hypothesis, neurons in layer III of somatosensory cortex have

23 11 extensive intracortical projections that allow them to communicate with neighboring neurons with similar receptive fields (Jones et al. 1978; Schwark and Jones 1989; Bernardo et al. 1990; Lund et al. 1993; Burton and Fabri 1995). Such connections allow SI neurons to receive sensory information from outside their receptive field, even though these subthreshold inputs are not evident until the local inhibitory circuits are antagonized (Dykes et al. 1984; Alloway et al. 1989; Alloway and Burton 1991; Kyriazi et al. 1996). Although intracortical connections probably are involved in reorganizing SI cortex after digit amputation, nerve transection, or other forms of sensory deprivation (Merzenich et al. 1984; Pons et al. 1991; Diamond et al. 1994; Fox 1994), their functional role during normal somatosensory processing remains unclear. One possible function for intracortical connections within SI cortex is to synchronize the activity of adjacent population of neurons during certain stimulus conditions. It is conceivable, for example, that intracortical connections might prime neighboring cortical populations to respond more effectively to a cutaneous stimulus that moves across the skin. To determine whether neuronal synchronization might have a role in coding somatosensory information, we compared the amount of synchronization present in spontaneous activity with that produced by cutaneous stimulation. Furthermore we also tested the possibility that a moving stimulus enhances synchronization in SI cortex more than a stationary stimulus.

24 Materials and Methods Four adult cats were used in this study and were treated according to National Institutes of Health guidelines for the use and care of laboratory animals. Most experimental procedures were described previously and are only briefly reported here (Johnson and Alloway 1994). Sterile operating techniques were used to expose SI cortex and to implant a stainless steel recording chamber onto the surrounding cranium. During this operation, a stainless steel bolt was attached to the occipital ridge to immobilize the animal s head during subsequent recording experiments. After implantation of the recording chamber, SI activity was recorded from each cat twice per week for 4 6 wks. During each recording session, the animal was intubated through the oral cavity and ventilated with a 2:1 gaseous mixture of nitrous oxide and oxygen containing % halothane. Heart rate and end-tidal CO 2 were monitored continuously, and body temperature was maintained at 37 C by a thermostatically controlled heating pad Cortical electrophysiology During each recording session, an array of 3 6 tungsten electrodes (2 5 MV; Frederick Haer, New Brunswick, ME) was used to record neuronal discharges in SI cortex. In virtually all experiments, the electrode arrays were configured to sample neurons separated by no more than 600 µm to ensure that all of the recorded neurons had overlapping receptive fields (Dykes and Gabor 1981). Three or four electrodes, arranged in a linear configuration (1 x 3 or 1 x 4;300-µm separation), were used in the initial experiments. In later experiments, a matrix of six electrodes (2 x 3; 250-µm separation) was used to record a larger number of neuron pairs simultaneously. The electrode array

25 13 entered the forearm representation of SI cortex located in the rostromedial bank of the coronal sulcus (Felleman et al. 1983). Electrodes penetrated the cortex at a 25 angle to the parasagittal plane and were advanced by a hydraulic microdrive until single neurons could be isolated on at least two or more electrodes. Recordings were made only from layers III or IV because the neurons in those layers are the most responsive to cutaneous stimulation (Johnson and Alloway 1996). Extracellular neuronal waveforms were displayed on an oscilloscope and converted into digital signals for off-line data analysis (DataWave Technologies, Broomfield, CO) Cutaneous stimulation Once neurons were isolated on multiple electrodes, their receptive fields (RFs) were mapped by manually stroking the hairy skin while listening to their neuronal discharges over an acoustic speaker. Most neurons recorded in this study were sensitive to hair movements and could be activated by jets of air that stimulated their RFs. Computer-controlled air jets were presented in blocks of 100 or 200 trials. Each trial was subdivided into three periods: a prestimulus period for recording spontaneous activity, a stimulation period that contained a series of stationary and/or moving air jets, and a poststimulus period. Neuronal activity was recorded during all three periods but was not recorded during intertrial intervals, which lasted 2 s. Prestimulus and poststimulus periods lasted 3 and 2 s, respectively. The duration of the stimulation period ranged from 2 to 6 s and depended on the number of air jets that were delivered. In the initial experiments, the stimulus period contained only stationary air jets or a moving air jet; in later experiments both types of air jets were presented within each trial. Stationary air jets

26 14 were delivered by three or four hollow tubes (1mm ID) that were aligned in a micromanipulator. The tubes were spaced at equal intervals, ranging from 5 to 20 mm, and were oriented orthogonally to the hairy skin surface. Each tube was connected to a four-channel manifold in which each channel s air flow was controlled by an electronic valve (Clippard ET-2 M). The electronic valve for each channel was controlled by a digital timer that was triggered by the data acquisition system (DataWave Technologies). Air pressure (20 psi) to the manifold was regulated by a needle valve in series with a pressure gauge. Previous work has shown that moving air jets activate discrete regions of the hairy skin without producing the lateral distortions caused by dragging a probe across the skin (Ray et al. 1985). Therefore we modified a Grass polygraph pen module to deliver moving air jets in a curvilinear trajectory. The ink pen of the polygraph module was replaced by a tube identical to those used for the stationary air jets except that the end of the tube was curved to direct a jet of air orthogonal to the sweeping motion of the tube. Air flow through the tube was controlled by an electronic valve as described for the stationary air jets. A waveform generator in series with a DC-coupled amplifier was used to produce constant velocity sawtooth movements of the air-jet tube. The waveform generator cycled at 0.5 or 1.0 Hz so that a moving jet of air traversed forward and backward across the skin for 1 or 2 s. The amplitude of the waveform generator was adjusted to produce a trajectory of movement that corresponded to the length of the RFs combined from all recording sites. Given the variability in RF sizes, the stimulus velocities for moving air jets ranged between 4 and 14 cm/s and averaged nearly 10 cm/s.

27 15 For each experiment, the moving air jet was positioned to pass over the same sites stimulated by the stationary air jets. In the initial experiments, stationary and moving air jets were presented in separate blocks of trials. In later experiments, a single block of trials was administered in which moving and stationary air jets were presented sequentially during each trial Analysis of neuronal responses Neuronal discharges were sorted on the basis of several parameters including spike width, spike amplitude, and time of maximum spike peak. The time of each neuronal discharge was recorded to within 0.1 ms, and time stamps from each group of sorted waveforms were used to generate summed peristimulus histograms (PSTHs) and cross-correlograms (CCGs). Binwidths for the PSTHs and CCGs were 25 and 0.5 ms, respectively. Cross-correlation analysis was used to characterize neuronal activity at one electrode as a function of neuronal activity recorded at a second electrode (Perkel et al. 1967). In a stimulus-based paradigm, an increase in correlated neuronal activity can be produced by stimulus coordination or may occur by chance due to the increased rate of neuronal discharges during peripheral stimulation. To estimate the magnitude of these effects, a linear shift predictor was subtracted from the raw CCG to produce a shiftcorrected CCG (Gerstein and Perkel 1972; Alloway et al. 1993; Johnson and Alloway 1996). For our analyses, the shift predictor was the mean of three CCGs calculated from pairing the first 97 of 100 reference responses (or 197 of 200 responses when 200 trials were administered) with subsequent target responses shifted by one, two, or three

28 16 stimulus trials. Using only 97 trial responses in a linear shift, rather than all 100 trial responses in a circular shift, avoided the pairing of responses having large time separations. Because stimulus-induced responses are not identical from one trial to the next, subtraction of the shift predictor may not remove all instances of stimulus coordination. Nonetheless, the shift predictor can detect many instances of stimulus coordination, especially at stimulus onset when neurons are most responsive and their response latencies are similar across trials. The shift predictor also was used because it represents a convenient tool for determining if correlated events are statistically significant. Because the shift predictor was based on independent spike trains (recorded in response to separate stimuli), the counts in each bin of the shift predictor were assumed to reflect a Poisson process and were used to calculate a Z score to evaluate the significance of values obtained in the shift-corrected CCG. The square root of each value in the shift predictor was multiplied by 1.96 to yield a 95% confidence limit (Aertsen et al. 1989). Peaks within the shift-corrected CCG that exceeded the 95% confidence limits on two or more contiguous bins were considered statistically significant (Aertsen et al. 1989; Gochin et al. 1989). CORRELATION COEFFICIENT. The correlation coefficient, ρ(τ 0 ), was calculated to indicate the proportion of discharges in the spike trains that were correlated (Abeles 1982). The formula for calculating the cross-correlation coefficient was adapted from Eggermont (1992): ρ(τ) = [ CE ] 2 2 {[ N ((N ) /T) ]* [ N ((N ) /T) ]} T T R R

29 17 where CE is the number of correlated events in the two tallest adjacent bins of a significant peak in the raw or shift-corrected CCG, T is time interval over which the CCG was calculated, and N T and N R represent the number of neuronal discharges recorded from the target and reference neurons during time T. SYNCHRONIZATION RATE. Because the correlation coefficient is independent of firing rate and does not indicate how often neuron pairs discharge simultaneously, we also calculated the rate of synchronized discharges from the raw and shift-corrected CCGs. For this parameter, the number of coincident events in the highest 2-ms peak was divided by the total recording time. Thus, the synchronization rate expresses the number of coincident events occurring per second. An interval of 2 ms was chosen for measuring synchronization rate because this duration encompasses most sharply synchronized events in local regions of SI cortex (Swadlow et al. 1998). PEAK HALF-WIDTH. We measured the peak half-widths of the shift-corrected CCGs to determine the amount of temporal variability among correlated discharges. Peak half-width was obtained by measuring the width of the CCG peak at half the height of its tallest bin. For spike trains having low rates of synchronized activity, peak half-width was difficult to measure because the bin heights were highly variable. For this reason, we ignored single 0.5-ms bins that dipped into a broader CCG peak. We also measured peak half-widths from both smoothed and unsmoothed CCGs. Smoothed CCGs were generated by averaging each bin in the unsmoothed CCG with its two adjacent bins. Although smoothing removes much of the variability in CCG peak, it may cause an increase in the width of CCG peaks, which consist of only one or two tall bins. Thus

30 18 smoothing was useful for measuring peak half-widths in CCGs based on low rates of spontaneous activity but was less accurate for measuring highly synchronized responses evoked by peripheral stimulation. Therefore we used smoothed CCGs to compare spontaneous and stimulus-induced synchronization but used unsmoothed CCGs to compare the degree of synchronization produced by stationary and moving air jets Autocorrelation analysis Autocorrelograms (ACGs) were constructed from spontaneous and stimulusinduced activity to detect oscillations among neurons showing coordinated responses in the shift-corrected CCGs. For this analysis, ACG peaks exceeding the expectancy value by at least 2 SDs (95% confidence limits) were considered statistically significant, and the ACG had to contain three or more significant peaks at regular temporal intervals to be classified as oscillating (Eggermont 1992). Each ACG was displayed over a time frame of at least ±125 ms Histology The final recording session for each animal was terminated by deeply anesthetizing the animal with an intravenous injection of pentobarbital sodium followed by an injection of 20 mg lidocaine and 1,000 USP units of heparin. The animal was transcardially perfused with 0.9% saline, neutral formalin, and 10% sucrose in formalin. The brain was removed and placed into 30% sucrose in formalin until it sank. The SI cortex was blocked, frozen, and cut into 50 µm coronal sections that were mounted onto chrom-alum-coated slides and stained with thionin.

31 19 Table 2.1 Spontaneous and stimulus-induced neuronal firing rates in SI cortex 2.3. Results Spontaneous Activity Preferred Direction Moving Air Jets Nonpreferred Direction We recorded extracellular discharges from ~700 SI neurons but excluded from analysis those neurons that did not respond to air-jet stimulation or that failed to discharge at least five times per stimulus trial. On the basis of these criteria, we acquired stimulus-induced responses from 544 SI neurons. In this sample, 263 neurons were stimulated by moving air jets in one block of trials and by stationary air jets in a separate blocks of trials; the remaining 281 neurons were stimulated by both types of air jets in a single block of trials. For this latter group, Table 2.1 displays the mean discharge rates recorded during spontaneous activity and in response to moving and stationary air jet stimulation. As Table 2.1 indicates, the SI neurons in this study exhibited low rates of spontaneous activity and were activated by stationary and moving air jets. Statistical comparison of the responses evoked from the most effective stationary site and those evoked by the preferred direction of movement indicates that these neurons were more responsive to moving stimulation (paired t = 6.25, P < ). Stationary Air Jets Most Effective We assessed the coordination of stimulus-induced neuronal activity in 880 pairs of SI neurons. This group includes neuron pairs in which stationary and moving air jets were both administered within the same block of trials (n = 577) as well as those in Least Effective Rates 1.64 ± ± ± ± ± 0.23 The results, expressed as means ± SE in discharges per second, are based on 281 neurons stimulated by interleaved moving and stationary airjets. which moving and stationary air jets were administered in separate blocks of trials (n =

32 20 303). Inspection of the shift-corrected CCGs indicated that 10% (n = 88) of these neuron pairs were synchronized during both spontaneous and stimulus-induced activity Modulation of SI synchronization by stationary air jets Stationary air-jet stimulation produced synchronization of SI activity in 87 neuron pairs. These data were obtained from experiments in which stationary air jets were delivered alone (n = 27) or were interleaved with moving air jets on each trial (n = 60). An example of SI synchronization during spontaneous activity and in responses to a series of stationary air jets is shown in Figure 2.1. In this case, neuron CC52a was excited by each of the three air jets aimed at the ventral forelimb, while neuron CC52b preferred air-jet stimulation at the more distal sites. The smoothed shift-corrected CCG compiled from spontaneous activity revealed weak synchronization in which the peak half-width lasted 15 ms and the correlation coefficient was only Cortical synchronization was enhanced noticeably during air-jet stimulation as indicated by the fact that the peak half-widths were narrower (2 6.5 ms) and the correlation coefficients were larger ( ) than those obtained during spontaneous activity. The effect of cutaneous stimulation on SI synchronization is illustrated by a pair of scatter plots, which show the distribution of correlation coefficients and peak halfwidths for 87 neuron pairs tested with stationary air jets (Figure 2.2). Because more than one RF site was stimulated for each neuron pair, each point in Figure 2.2 represents the air-jet response having the largest correlation coefficient. Compared with spontaneous activity, air-jet stimulation caused a larger proportion of SI cortical activity to become

33 21 A B 200 CC52a1 CC52 a1 CC52 b1 Number of Events CC52b Spontaneous Jet 1 Jet 2 Jet s C Correlated Events ms ms 6.5 ms 4.5 ms Time (ms) Figure 2.1 Responses to stationary air jets by a pair of primary (SI) somatosensory cortex neurons that had overlapping receptive fields (RFs) and were separated by a distance of 300 µm. A: outline drawing of the forepaw showing the RF boundaries for neurons CC52a and CC52b and the sites (?) that were stimulated by a series of stationary air jets. B: peristimulus histograms (PSTHs) of the responses evoked by stimulating the sites indicated in A; binwidth, 25 ms. Mean extracellular waveforms of neurons CC52a and CC52b are shown (inset); bars, 1.0 ms and 50 mv. C: smoothed, shift-corrected cross-correlograms (CCGs) show the pattern of correlated discharges occurring spontaneously and in response to stationary airjet stimulation. Magnitude of the correlation coefficients and peak half-widths are indicated on each CCG. dotted line, 95% confidence limits; binwidth, 0.5 ms.

34 22 A 0.1 Correlation Coefficients B 100 Peak Halfwidths (ms) Stationary Airjet Response 0.01 Stationary Airjet Response Spontaneous Activity Spontaneous Activity Figure 2.2 Scatter plots comparing the parameters of cortical synchronization for 87 neuron pairs that had correlated discharges during spontaneous activity and stationary airjet stimulation. Left and right: illustrations of correlation coefficients and peak half-widths, respectively, which were obtained from smoothed, shift-corrected CCGs. Because multiple sites in the RFs were stimulated, each data point is based on the response obtained from the stimulus site with the largest correlation coefficient. The dashed line indicates where correlation coefficients or peak half-widths were identical during spontaneous and stimulus-induced activity. Arrows show data obtained from the pair of neurons illustrated in Figure 2.1. synchronized. Thus the mean correlation coefficient increased from ± (mean ± SE) during spontaneous activity to ± during stationary air-jet stimulation (paired t = 7.99; P < ). Cutaneous stimulation also reduced the temporal variability of correlated discharges in the local network as indicated by the decline in peak half-widths from a mean of 8.17 ± 1.37 ms during spontaneous activity to a mean of 3.93 ± 0.41 ms during stationary air-jet stimulation (paired t = 3.06; P < 0.002).

35 Modulation of SI synchronization by moving air jets Cutaneous stimulation with a moving air jet caused synchronization in 88 neuron pairs. These data represent the combined results from experiments in which moving air jets were administered alone (n = 28) as well as experiments in which moving and stationary air jets were interleaved on each trial (n = 60). Figure 2.3 illustrates the effects of a moving air jet on the same pair of neurons the responses of which to stationary air jets were shown in Figure 2.1. The mean extracellular waveforms in Figures 2.1 and 2.3 show that both neurons remained well isolated, although the waveform for neuron CC52a had increased in amplitude since the time the data in Figure 2.1 were recorded. The spontaneous activity of neuron CC52a also had declined slightly, but both neurons continued to respond vigorously to air jets moving distally or proximally across their RFs. Although cross-correlation analysis indicated little change in the half-widths of the CCG peaks obtained during spontaneous activity (4 ms) or in response to a moving air jet (2 ms), the change in the proportion of correlated discharges was quite striking. The smoothed shift-corrected CCG for spontaneous-activity contained a small peak near time 0, which had a correlation coefficient of During stimulation with a moving air jet, the correlation coefficient increased to during movement in the preferred (forward) direction but declined to when the air jet moved in the nonpreferred (reverse) direction even though both neurons showed similar rates of activity (8.85 vs spikes/s for CC52-a1 and 60.0 vs spikes/s for CC52-b1) in both directions. The lack of a relationship between firing rate and correlation coefficient was underscored by the fact that the correlation

36 24 A B 100 CC52 a CC52 a1 CC52 b1 Number of Events CC52 b Spontaneous Forward Reverse 7.0 s C Correlated Events ms ms ms Time (ms) Figure 2.3 Responses of the same pair of SI neurons shown in Figure 2.1 but during subsequent stimulation with a moving air jet. A. Dotted line indicates trajectory of the air jet as it moved between points 1 and 2. B: PSTHs show that both neurons responded vigorously to an air jet moving in the forward or reverse directions. C: smoothed, shift-corrected CCGs showing the pattern of correlated discharges during spontaneous activity and in response to air-jet movements in the forward and reverse directions.

37 25 coefficient during movement in the nonpreferred direction (0.0151) was lower than during spontaneous activity (0.0157) even though the discharge rates of both neurons was several times higher during cutaneous stimulation. This was not an isolated case as 17% (n = 15) of the neuron pairs showed directional preferences in synchronization without showing corresponding changes in their underlying rate of activity. Furthermore, smoothing the shift-corrected CCGs was not a factor in these cases because the unsmoothed CCGs displayed the same degree of directional preferences in their correlation coefficients. The effects of moving air jets on cortical synchronization are summarized for all 88 neuron pairs in Figure 2.4. The scatter plots in this figure show the distribution of A 0.1 Correlation Coefficients B 100 Peak Halfwidths (ms) Moving Airjet Response 0.01 Moving Airjet Response Spontaneous Activity Spontaneous Activity Figure 2.4 Scatter plots comparing the parameters of cortical synchronization for 88 neuron pairs that had correlated discharges during spontaneous activity and in response to moving air-jet stimulation. Left and right: illustrations of correlation coefficients and peak half-widths, respectively, which were obtained from the smoothed, shift-corrected CCGs. Both scatter plots show only the data obtained from responses to air jets moving in the direction evoking the largest correlation coefficient. Arrows indicate data illustrated in Figure 2.3.

38 26 correlation coefficients and peak half-widths obtained during spontaneous activity and moving air-jet stimulation. Because the pattern of coordination often differed when the air jet moved in opposite directions, each data point represents the air-jet response having the largest correlation coefficient. Compared with spontaneous activity, moving air jets produced a substantial increase in cortical synchronization. Thus the correlation coefficients increased from a mean of ± during spontaneous activity to ± during air-jet movements in the preferred direction. A matched-sample t-test indicated that these differences were highly significant (paired t = 4.63; P < ). A similar comparison of mean peak half-widths obtained during spontaneous activity (7.03 ± 1.31 ms) and during air-jet movement (3.03 ± 0.30 ms) indicated that moving air jets caused a decrease in the temporal variability of correlated activity (paired t = 3.11; P < 0.002) Comparison of synchronization induced by stationary and moving air jets We compared the effects of stationary and moving air-jet stimulation on 60 neuron pairs that showed significant levels of cortical synchronization during stationary and moving air-jet stimulation. This analysis was conducted only on neuron pairs in which stationary and moving air jets were both administered within the same block of trials. This restriction was necessary because many spike trains showed clear signs of nonstationarity when different blocks of trials were compared (see Figures 2.1 and 2.3). In addition to analyzing single neurons, we also analyzed multiunit responses recorded across pairs of electrodes. Multiunit activity was analyzed because many electrodes recorded two or three distinguishable waveforms, but cross-correlation

39 27 analysis usually failed to detect coordination among any of the single neuron pairs even though their PSTHs were highly similar. Cross-correlation analysis of multiunit responses revealed significant levels of correlated activity across 79 pairs of electrodes. A representative example comparing the effects of stationary and moving air-jet stimulation on SI synchronization is presented in Figure 2.5. In this case, three distinct waveforms were recorded from one electrode (CC69a) and two neuronal waveforms were recorded simultaneously by an electrode located 250 µm away (CC69b) to yield a total of six neuron pairs. Cross-correlation analysis revealed substantial amounts of synchronization in the multiunit responses and in four of the six single neuron pairs. The shift-corrected CCGs obtained from the multiunit responses contained tall peaks in which correlation coefficients were largest during moving air-jet stimulation (0.120) and smallest during spontaneous activity (0.104). The half-widths of these peaks showed that the relative timing of correlated activity was less variable during moving air-jet stimulation (0.5 ms) than during spontaneous activity (1.5 ms) or during stationary air-jet stimulation (2.5 ms). Comparison of the multiunit CCGs with those obtained from single neurons revealed noticeable variability in the coordination of specific pairs of neurons. In one pair of neurons (a3 and b2), for example, each of the CCGs obtained from spontaneous and stimulus-induced activity contained a prominent peak centered around time 0. In another pair of neurons (a1 and b1), correlated activity was barely detected during the moving air-jet response, whereas the stationary air jets produced a broad peak of correlated activity that was located 1 2 ms to the right of time 0.

40 28 Figure 2.5. Comparison of moving and stationary air jets on neuronal synchronization in SI cortex. A: outline drawing of the forepaw showing the overlapping RF boundaries for the multiunit activity recorded by electrodes CC69a and CC69b. Moving air jet pursued a curvilinear trajectory between points 1 and 2 (curved line) and was followed by a series of stationary air jets directed at sites A and B (filled circles). B: PSTHs showing the responses of multiple neurons recorded by a pair of electrodes (CC69a and CC69b) separated by 250 µm. Binwidths, 25 ms. C: shift-corrected CCGs showing the patterns of correlated discharges that occurred during spontaneous activity and in response to moving and stationary air jets. Top: unsmoothed CCGs that were obtained from an analysis of multiunit activity. Bottom: CCGs obtained from an analysis of 2 pairs of single neurons (a3 and b2, a1 and b1) during spontaneous activity (smoothed) or in response to stationary and moving air jets (unsmoothed). Each CCG portrays changes in the activity of CC69a as a function of CC69b activity at time 0. Binwidths, 0.5 ms.

41 29 Although the correlation coefficients for CC69 appeared similar during both moving and stationary air-jet responses, mean firing rates were higher during air-jet movement. This difference was not apparent from CCGs generated from the complete stimulus period because the stationary air jets lasted 1,000 ms whereas the moving air jets lasted only 500 ms in each direction. Furthermore although each stationary air jet was aimed at overlapping regions of the neuron s RFs, the moving air jets traversed the entire length of each RF, and this meant that the beginning and end of each sweep of the moving air jet was ineffective for activating both neurons. Therefore to fully appreciate any differences in the rate of synchronized activity produced by moving and stationary air jets, it is necessary to examine the rate of correlated activity produced during equivalent time periods when both types of air jets stimulate overlapping portions of the RFs. Because moving air-jet responses were largest in the midst of the sweep, we conducted cross-correlation analysis on the activity occurring in the middle portion ( ms) of an air jet moving in the preferred direction. Furthermore we also conducted cross-correlation analysis on equivalent 300-ms periods at the beginning, middle, and end of the best stationary air jets. Figure 2.6 illustrates the differences produced by moving and stationary air jets by presenting the raw and shift-corrected CCGs calculated from 300 ms periods for the same multiunit responses shown in Figure 2.5. The differential effects of moving and stationary air jets on the rate of synchronized activity is clear from comparing the amplitude of the CCG peaks in Figure 2.6. Whereas stationary air jets produced relatively short and broad CCG peaks for each 300-ms period, the moving airjet produced a much taller peak of correlated activity. In fact, most of the correlated

42 30 Figure 2.6 Comparison of moving and stationary air jets on the synchronization rate of the same multiunit responses shown in Figure 2.5. All raw CCGs (top) and shift-corrected CCGs (bottom) were constructed from 300-ms time periods obtained during the middle of the moving jet ( ms) or from the beginning ( ms), middle ( ms), or end ( ms) of stationary jet B. All CCGs are unsmoothed. Correlation coefficient was largest during the moving air jet (0.127) and declined progressively during each period of the stationary air jet (0.117 to 0.096). When the raw CCGs were analyzed, synchronization rates were highest for the moving air jet (44.6 coincident events/s) and were lower during all periods of the stationary airjet ( coincident events/s). The same pattern also was present when the shift-corrected CCGs were analyzed. Binwidths, 0.5 ms. activity produced by the moving air jet occurred precisely at time 0 in the raw and shiftcorrected CCGs, and the rate of synchronization during the moving air jet (44.6 coincident events/s) was substantially higher than that produced by the stationary air jet (ranging from 26.4 to 16.8 coincident events/s). A comparison of the correlation coefficients showed that the proportion of synchronized activity was higher during the moving air jet (0.127) than during the initial period (0.117) or in any subsequent part of the stationary air jet ( ).

43 31 Figure 2.7 Comparison of stationary and moving airjets on synchronization rate. Each bar indicates the mean rate of coincident events during experiments in which stationary and moving airjets were both presented on each trial. These data were obtained from raw CCGs (top) and shift-corrected CCGs (bottom) constructed from discharges in the 3,000-ms prestimulus period (spontaneous), the entire stationary air-jet period (stationary 1,000 ms), the best 300-ms period during the stationary air jet (stationary 300 ms), and the middle 300-ms period of the moving air jet (moving). All CCGs were unsmoothed except for those constructed from the spontaneous activity of single units. Brackets indicate SE. *** indicates responses to the stationary air jets that were significantly different from the moving airjet response (matched-sample t-tests; ***P < 0.001). The effects of stationary and moving air jets on the mean rate of synchronization are summarized for single- and multi-unit responses in Figure 2.7. For stationary air jets, we measured synchronization rate for the complete stimulation period (1,000 ms) and from the best 300-ms period occurring at the beginning, middle, or end of the air jet as shown in Figure 2.6. In 90% of the cases, synchronization rates were highest during the first 300-ms period and declined progressively during the remaining periods. For moving air jets, synchronization rates were calculated for the 300-ms period in the middle of the sweep in the preferred direction as shown in Figure 2.6. We did not calculate the rate of

44 32 synchronization for the entire 500-ms period because many neurons were not activated by the onset or end of each sweep. As shown by Figure 2.7, regardless of whether we examined the raw or shift-corrected CCGs, moving air jets produced higher rates of synchronized activity than stationary air jets for both the single- and multiunit responses (paired t = 3.56, P < for all comparisons). To compare the proportion of activity synchronized by stationary and moving air jets, we calculated the correlation coefficients for the single- and multiunit responses that had the highest synchronization rates. As Figure 2.8 indicates, mean correlation coefficients for multiunit responses were substantially larger than those obtained from pairs of single neurons. This result is consistent with other evidence that correlation Figure 2.8 Comparison of the proportion of activity that was correlated during stationary and moving air jets. Each bar indicates the mean correlation coefficient obtained from the same CCGs represented in Figure 2.7. Brackets indicate SE. Asterisks indicate responses to the stationary air jets that were significantly different from the moving air-jet responses (matched-sample t-tests; ***P < 0.001).

45 33 coefficients are larger when multiunit responses are analyzed because there are more opportunities for detecting correlated discharges that occur among subsets of different neuron pairs (Bedenbaugh and Gerstein 1997). As shown by Figure 2.8, the raw CCGs for both single- and multiunit activity had higher correlation coefficients during the moving air jets than during the stationary air jets (paired t = 4.65, P < for multiple neurons; paired t = 3.47, P < for single neurons). Analysis of the shift-corrected CCGs, however, failed to reveal any significant difference in the proportion of correlated activity produced by moving or stationary air jets among pairs of single or multiple neurons. Thus subtraction of the shiftpredictor caused considerable reduction in the correlation coefficients, but the resulting levels remained above those obtained during spontaneous activity. A comparison of the mean peak half-widths obtained from the shift-corrected CCGs for single and multiple neurons is shown in Figure 2.9. We did not measure peak Figure 2.9. Comparison of stationary and moving air jets on the timing of synchronized activity. Each bar indicates the mean peak half-width obtained from the shift-corrected CCGs represented in Figures 2.7 and 2.8. Except for the CCGs representing the spontaneous coordination of single units, all other CCGs were unsmoothed. Brackets indicate SE. Asterisks indicate stationary airjet responses that were significantly different from the moving airjet response (matched-sample t-tests; *P < 0.05)

46 34 Table 2.2 Probability of synchronization across SI recording sites separated by varying distances Distance Separating Electrodes, µm Total Multiple Neuron Pairs Pairs Recorded Significant Pairs Percentage, % Single neuron pairs Pairs recorded Significant pairs Percentage, % half-widths for the raw CCGs because we frequently found that half the height of the tallest peak was within the background level of correlated activity which rendered this parameter meaningless. Consistent with the results shown previously in Figures 2.2 and 2.4, both moving and stationary air jets produced substantial decreases in CCG peak halfwidths when compared with spontaneous activity. Peak half-widths for the stationary airjet responses were larger when the entire 1,000-ms period was analyzed, and this result parallels a trend seen in 90% of the single- and multiunit responses in which the rate and proportion of synchronized activity gradually declines during successive 300-ms periods of the stationary air jet (see Figures ). Analysis of the multiunit responses showed that moving air jets produced slightly less variability in the timing of correlated activity than stationary air jets (paired t = 2.04, P < 0.05 for comparison with the best 300-ms period). A similar comparison of single neuron pairs, however, failed to detect significant differences in the peak half-widths produced by moving and stationary air jets (paired t = 2.005, P < for comparison with the entire 1,000-ms period).

47 35 Figure 2.10 Changes in the rate of synchronization produced by stationary and moving air jets as a function of distance between electrodes. Each bar indicates the mean synchronization rate obtained from analyzing raw (top) and shift-corrected (bottom) CCGs generated from multiunit responses in the 3,000-ms prestimulus period (spontaneous) and in the 300-ms periods having the highest synchronization rate during the stationary and moving air jets. Number of synchronized CCG patterns detected at each distance varied: 250 µm, n = 30; 300 µm, n = 19; 354 µm, n = 19; 600 µm, n = 9. Brackets indicate SE. Asterisks indicate stationary air-jet responses that were significantly different from the moving airjet response at a given electrode separation (matched-sample t-tests; *P < 0.05, **P < 0.01, ***P < 0.001) Effects of electrode separation on stimulus-induced cortical synchronization There was considerable RF overlap when neurons were separated by only µm, but the amount of overlap declined with increasing distance and was much less than 50% of the combined RFs for neurons separated by 600 µm. This was consistent with other reports showing that RF overlap varies systematically with cortical separation (Sur et al. 1980; Dykes and Gabor 1981; Alloway and Burton 1985). Unless objective criteria are used, however, RF boundaries are difficult to define and may vary according to experimenter bias. We also found that obtaining precise RFs for each neuron was problematic when two or more neurons were recorded simultaneously from the same

48 36 Figure 2.11 Changes in the proportion of synchronized activity produced by stationary and moving air jets as a function of the distance between electrodes. Each bar indicates the mean correlation coefficient obtained from the same CCGs represented in Figure Brackets indicate SE. Asterisks indicate stationary air-jet responses that were significantly different from the moving air-jet response at a given electrode separation (matched-sample t-tests; **P < 0.01, ***P < 0.001). electrode. For these reasons, we analyzed cortical synchronization as a function of the distance between pairs of electrodes because this parameter was measured easily and, with distances of = 600 µm, appeared to be correlated with RF overlap. Table 2.2 indicates that the probability of detecting cortical synchronization declined with increasing distance between recording sites. Because this trend was less evident among single neuron pairs, we analyzed only the CCGs of multiunit responses to determine how cortical synchronization varied with electrode separation. For both the raw and shift-corrected CCGs, electrode separation had a significant effect on synchronization rate (F = 72.4, P < for raw CCGs; F = 66.5, P < for shiftcorrected CCGs). Synchronization rates were higher for moving air jets than for stationary air jets at each of the electrode separations, and the highest synchronization rates were recorded at separations of 300 µm (Figure 2.10). Unexpectedly,

49 37 synchronization rates were higher for neurons separated by 600 µm than for neurons separated by 250 or 354 µm. Electrode separation also had a significant effect on correlation coefficients, but this effect was more evident in the raw CCGs than in the shift-corrected CCGs (F = 32.7, P < for raw CCGs; F = 5.5, P < 0.01 for shiftcorrected CCGs). Analysis of correlation coefficients from the raw CCGs also revealed a propensity for greater amounts of synchronization at 300-µm increments (Figure 2.11). When the shift-corrected CCGs were analyzed, however, the proportion of correlated activity was highest at 300-µm intervals, but there was little difference in this parameter at separations of 250, 354, or 600 µm. Electrode separation also had a significant effect on the temporal variability of synchronized discharges (F = 23.8, P < 0.001), but this effect was more apparent for spontaneous than for stimulus-induced synchronization (Figure 2.12). Thus the mean half-width of CCG peaks constructed from spontaneous activity increased from =6 ms at the short intervals ( µm) to nearly 15 ms at intervals of 600 µm. By contrast, cortical synchronization produced by moving or Figure 2.12 Changes in the timing of synchronized activity produced by stationary and moving air jets as a function of distance between electrodes. Each bar indicates the mean peak half-width obtained from the shift-corrected CCGs represented in Figures 2.10 and Brackets indicate SE

50 38 stationary air jets had little temporal variability and mean peak half-widths remained <2.40 ms at the shorter intervals ( µm). At separations of 600 µm, however, peak half-widths increased to a mean value of 2.77 ± 1.13 ms for the moving air jets and 4.33 ± 0.26 ms for the stationary air jets (paired t = 2.35, P < 0.055) Lack of oscillations during stimulus-induced synchronization Many reports indicate that neuronal oscillations may play an essential role for synchronizing activity in segregated cortical areas (Eckhorn et al. 1988; Gray et al. 1989; Gray et al. 1992; Bressler et al. 1993; Steriade et al. 1994; Konig et al. 1995; Gray and McCormick 1996; Murthy and Fetz 1996; Murthy and Fetz 1996). Therefore, we applied autocorrelation analysis to the multiunit responses to determine if stimulus-induced synchronization was dependent on oscillatory activity. In this analysis, oscillatory activity was considered to be present if the ACG contained three or more peaks which exceeded the expectancy level by 2 SDs at regular temporal intervals (Eggermont 1992). Only five recording sites showed clear cases of oscillatory activity during air-jet stimulation, but two of these responses were identical to oscillatory patterns that appeared spontaneously. None of the remaining multiunit responses contained any clear patterns of oscillation, and this is consistent with other reports indicating that oscillations are not necessary to synchronize cortical neurons separated by less than 2 mm (Konig et al. 1995; Swadlow et al. 1998). Figure 2.13 illustrates some examples of multiunit responses that were synchronized during moving and stationary air-jet stimulation yet failed to display any oscillations.

51 39 Figure 2.13 Lack of oscillations among cortical neurons showing stimulus-induced synchronization. Multiunit responses were recorded at 2 electrodes separated by 600 (CC74)or 250mm (CC88). Top: shiftcorrected CCGs based on the multiunit responses to a 500-ms moving air jet or a 1,000-ms stationary air jet. Bottom: autocorrelograms (ACGs) based on the responses recorded at each electrode. Numbers on each ACG represent the number of discharges recorded during each air-jet stimulation period. Binwidths, 1.0 ms.

52 Discussion The results of this study demonstrate that neuronal synchronization is an important part of the cortical response to tactile stimulation. Consistent with the view that neuronal synchronization is a potential mechanism for coding certain aspects of sensory stimuli, our results indicate that correlated activity in somatosensory cortex may supplement the changes in firing rate that code intensity and other attributes of a cutaneous stimulus. Although many studies have shown that neuronal synchronization might play a role in visual perception, this is one of the first studies to show the potential utility of stimulus-induced synchrony in somatosensory cortex Anatomic factors affecting synchrony in SI cortex The probability of detecting synchronized responses to a discrete air jet was highest for SI neurons located across an interval of 300 µm and was substantially lower for neurons separated by µm (see Table 2.2). One factor that appears to be related to this trend is the degree of RF overlap. Neurons separated by µm share the majority of their RFs, whereas neurons separated by µm share only a small portion of their RFs. This observation confirms reports showing that RF overlap in somatosensory cortex is related inversely to the distance intervening between neurons (Sur et al. 1980; Dykes and Gabor 1981; Alloway and Burton 1985). Those studies also found that neurons separated by µm had nonoverlapping RFs that represented adjacent skin regions. Because we rarely recorded responses from sites separated by >600 µm, we do not know if a discrete moving stimulus can synchronize SI populations that are spaced more widely.

53 41 The incidence and strength of synchronization between neighboring parts of SI cortex also appears to be related to anatomic factors. Like other cortical areas, pyramidal neurons in SI cortex, especially those in layer III, give rise to axonal collaterals that have extensive contacts with other neurons in the vicinity of the soma as well as neurons located more distantly (DeFelipe et al. 1986; Schwark and Jones 1989; Bernardo et al. 1990; Juliano et al. 1990; Lund et al. 1993; Burton and Fabri 1995). Although the spatial distribution of intracortical connections probably is related to RF overlap, any discontinuities in this distribution might explain why the incidence and rate of synchronization were higher for neurons separated by 600 µm than for neurons separated by 250, 354, 500, or 560 µm. Intracortical connections in striate cortex, for example, cluster at regular intervals (Gilbert 1992), and some evidence indicates that focal collateralizations also may occur at regular intervals in SI cortex (DeFelipe et al. 1986). Common inputs from thalamocortical projections are also likely to play a major role in synchronizing adjacent groups of SI neurons. Thalamocortical relay neurons have axon collaterals that terminate in multiple patches of SI cortex and may span a distance of =600 µm (Landry and Deschenes 1981; Snow et al. 1988; Garraghty and Sur 1990). Consistent with this finding, our CCG peaks usually straddled time 0, a coordination pattern that suggests the presence of common inputs from a third source (Perkel et al. 1967; Fetz et al. 1991) Synchronization in SI cortex and sensory coding Compared with spontaneous activity, both stationary and moving air jets caused substantial increases in the rate, proportion, and temporal precision of synchronized

54 42 activity in local regions of SI cortex. Although both types of stimuli produced large increases in the rate of synchronized activity, moving air jets were significantly more effective than stationary air jets in boosting this parameter. Furthermore the increased rate of synchronized activity produced during moving air-jet stimulation was not just the result of an increase in neuronal firing rate but was accompanied by significant increases in the proportion of correlated activity as measured by the correlation coefficients for the raw CCGs. Finally, differences in the rate of synchronization produced by stationary and moving air jets were most prominent among neural populations that were separated by 600 µm and thus had minimal RF overlap. This finding is important for sensory coding because the stationary and moving air jets were identical with respect to the skin area that was stimulated at any moment in time. Whereas a previous study found that a discrete tactile stimulus can synchronize SI cortical neurons separated by =500 µm (Metherate and Dykes 1985), our study extends those results by suggesting that synchronization may occur over wider regions of SI cortex if the stimulus sequentially activates groups of neurons representing contiguous skin regions. Some evidence suggests that sensory stimulation evokes recurrent excitation among neighboring cortical populations that may, under certain conditions, interact with incoming thalamocortical activity to enhance cortical synchronization (Douglas et al. 1995). On the basis of the differences in firing rate and proportion of correlated discharges produced by moving and stationary stimuli, we believe that moving stimuli are more effective than stationary stimuli in promoting cooperativity among related thalamocortical and corticocortical networks. In our view, moving stimuli sequentially

55 43 recruit neighboring populations of thalamocortical neurons that project to the subliminal fringe of excitation surrounding the cortical area activated in the preceding time frame. Compared with a stationary stimulus, a moving stimulus evokes more cortical excitation and continuously activates neighboring populations of thalamocortical and corticocortical networks that are strongly interconnected. One consequence of this appears to be a tremendous increase in the rate of highly synchronized activity in neighboring regions of cortex and suggests that local regions of cortex are wired to become synchronized when the same stimulus activates neighboring parts of this network. We speculate that the cortical area over which a single moving stimulus may cause neuronal synchronization probably is related to the speed of stimulus motion and the time period over which recurrent excitation persists. In any case, our findings are consistent with the view that synchronization is a plausible mechanism for linking adjacent cortical populations and suggest that shifts in highly synchronized activity from one cortical region to the next is an important neural correlate of the sensation of movement produced by a single moving stimulus. We also obtained preliminary evidence suggesting that neuronal synchronization in SI cortex can signal more specific attributes of a cutaneous stimulus. Thus 15 of our neuron pairs were strongly synchronized by air jets moving in one direction but not the other even though the underlying rate of activity was similar for both directions of movement. Although earlier studies indicate that some SI neurons are directionally sensitive (Whitsel et al. 1972; Warren et al. 1986; Ruiz et al. 1995), those studies did not analyze whether groups of such neurons become synchronized or whether

56 44 synchronization might code stimulus direction independent of changes in firing rate. Although our results suggest that synchronization might play a role in coding direction of movement, we only tested stimuli that moved back and forth for one stimulus cycle. Hence, in these cases we do not know whether synchronization might vary with the level of adaptation, the direction of the initial stimulus, or other factors Parallels with other sensory systems The presence of synchronized activity within distributed populations of cortical neurons has been investigated in many sensory systems because of its theoretical importance as a potential coding mechanism (Konig and Engel 1995), and many of the findings in those studies resemble our results in SI cortex. In auditory cortex, for example, synchronization among local groups of neurons is substantially greater during sound stimulation than during periods of spontaneous activity (Dickson and Gerstein 1974; Frostig et al. 1983; Eggermont 1994; decharms and Merzenich 1996). In addition, neighboring neurons in auditory cortex display stimulus-induced interactions that have narrower CCG peak widths and larger correlation coefficients than neuron pairs that are more widely separated (Eggermont and Smith 1996). Finally, for a significant fraction of neurons in auditory cortex, synchronization is sensitive to the direction of sound movement (Ahissar et al. 1992). Just as we have observed that synchronization in SI cortex is more likely for neurons sharing similar RF properties, a large body of data indicate that synchronization in visual cortex is governed by similar principles (Singer and Gray 1995). Thus local populations of striate neurons are more likely to become synchronized when they have

57 45 similar response properties and overlapping RFs (Toyama et al. 1981; Toyama et al. 1981; Gray and Singer 1989). Consistent with the Gestalt criteria for visual perception, adjacent populations of neurons in striate cortex are more likely to be synchronized if they have similar orientation and directional preferences (Ts'o et al. 1986) and represent collinear portions of the visual field (Gray et al. 1989). These findings indicate that the spatial continuity of a visual stimulus is important for organizing striate cortical neurons into functional assemblies (Singer and Gray 1995) Comparisons of multiple and single neuron responses Many laboratories have analyzed multiunit activity to reveal cortical synchronization during sensory stimulation. Cross-correlation analysis of multiunit activity has been used widely to demonstrate that neuronal synchronization is a potential coding mechanism in visual and auditory cortex (Eckhorn et al. 1988; Gray and Singer 1989; Engel et al. 1990; Engel et al. 1991; Gray et al. 1992; decharms and Merzenich 1996). In addition, temporal analysis of local field potentials also has been used to determine whether segregated populations of neurons become synchronized under various sensory or behavioral conditions (Gray and Singer 1989; Engel et al. 1990; Bressler et al. 1993; Sanes and Donoghue 1993; MacKay and Mendonca 1995; Murthy and Fetz 1996). Some investigators have concluded that cross-correlation analysis of multiunit activity is more sensitive for detecting cortical synchronization than analysis of single neuron pairs (decharms and Merzenich 1996; Bedenbaugh and Gerstein 1997). We agree with this view because synchronized activity was apparent in only 10% of our single

58 46 neuron pairs, yet appeared among 64% of our electrode pairs when multiunit responses were analyzed. Although the proportion of synchronized neuron pairs varied as a function of electrode separation, the ability to detect cortical synchronization over longer distances was improved greatly when multiunit responses were analyzed. Finally, small differences in the effects of stationary and moving air jets on mean peak half-widths were detected by our analysis of multiunit responses but not by a similar analysis of single neuron pairs. Although cross-correlation analysis of multiunit activity appears to be more sensitive for detecting neuronal coordination, the data acquired with this technique must be interpreted carefully (Bedenbaugh and Gerstein 1997). A potential problem with comparing multiunit responses with particular stimuli concerns the recruitment of different sets of neurons. If stationary and moving air jets activate different groups of neurons, any change in correlated activity produced by these stimuli might reflect a sampling difference rather than a change in functional connectivity. Moving air jets evoked higher rates of activity than stationary air jets (see Table 2.1), and it could be argued that a moving stimulus recruits more SI neurons than a stationary air jet. There are two reasons why this possibility does not explain the differences that we observed. First, we noted the characteristics of the extracellular waveforms recorded during each stimulus and did not observe differences in the shape, amplitude, or width of discharges evoked by moving and stationary stimuli air jets administered in the same block of trials. Second, a comparison of the responses to stationary and moving air jets showed that the CCG peak half-widths were slightly narrower during stimulus movement (see Figure 2.9). If moving

59 47 air jets activated a larger population of cortical neurons, then this should have produced an increase, not a decrease, in the temporal variability of their coincident discharges Interpretation of raw and shift-corrected CCGs In this study we presented data from both the raw and shift-corrected CCGs for a variety of reasons. First, raw CCGs represent the actual patterns of synchronized activity that are available to the organism for sensory perception and discrimination. Second, our stimuli are relatively long in duration ( ms), and responses to moving and stationary air jets can be highly variable from one trial to another. Thus it could be argued that subtraction of the shift predictor in our paradigm does not really remove enough stimulus-coordinated events to portray accurately the amount of correlated activity mediated by neural connections. Finally, we wished to determine whether the raw and shift-corrected CCGs revealed similar patterns in the relative amounts of synchronization produced by moving and stationary air jets. We found that moving air jets produced higher synchronization rates than stationary air jets and that this difference was present in both the raw and shift-corrected CCGs. By contrast, correlation coefficients were significantly higher for the moving air-jet responses when the raw CCGs were analyzed, but this difference disappeared when the shift predictor was subtracted. It is not immediately obvious why the correlation coefficients produced by moving and stationary air jets should be different in the raw but not in the shift-corrected CCGs. Any plausible explanation must consider what the shift predictor represents in these experiments. Consistent with the flat appearance of the 95% confidence limits, which are derived from the shift predictor, we did not observe any prominent peaks in the

60 48 shift predictor around time 0. Instead, the shift predictor was either completely flat or, in a few instances, contained relatively broad elevations that gradually tapered away from time 0. The lack of a prominent peak suggests that events in the shift predictor do not reflect temporal characteristics of the stimulus but could reflect correlations due to chance. In cases where neurons are not interconnected and do not share any common inputs, the shift predictor accurately indicates the probability of chance correlations, and this value is determined largely by the rate of activity in the recorded neurons. Because discharge rates were significantly greater during moving air jets than during stationary air jets, it could be argued that correlations due to chance are disproportionately greater for the responses to moving air jets. In cases where neurons are likely to be interconnected or to share common inputs, however, the coincident events subtracted from the raw CCG are likely to represent a combination of chance correlations and correlations produced by direct neuronal interactions or common inputs. In experiments such as ours, in which the neurons share RFs and are likely to share thalamocortical inputs, the differential effects of stationary and moving stimuli on synchronization rate may reflect true differences in the proportion of correlated activity produced by neuronal connections. Hence the larger correlation coefficients observed in the raw CCGs during moving air-jet stimulation suggest that this stimulus enhances the cooperativity of thalamocortical projections to common postsynaptic targets.

61 49 Chapter 3. Long-Range Cortical Synchronization Without Concomitant Oscillations in the Somatosensory System 3.1. Introduction A major issue in systems neuroscience concerns whether cortical synchronization is a mechanism for linking distributed populations of neurons that represent distinct attributes or features of a unitary stimulus (von der Malsburg 1994; Singer and Gray 1995; Singer et al. 1997). Although considerable controversy surrounds the view that cortical synchronization solves some of the coding problems associated with sensory perception (Bernardo et al. 1990; Ghose and Freeman 1992; Gray 1999; Shadlen and Movshon 1999; Singer 1999), the data indicating that sensory stimulation evokes correlated activity in local regions of cortex are indisputable. In primary visual cortex, for example, adjacent groups of neurons with similar stimulus preferences exhibit correlated activity in response to an optimal stimulus that appears in their receptive fields (Ts'o et al. 1986; Gray et al. 1989; Engel et al. 1991; Livingstone 1996). Furthermore, a growing body of evidence in other sensory systems demonstrates that cortical populations with similar response properties become synchronized during certain stimulus conditions (Dickson and Gerstein 1974; Metherate and Dykes 1985; Ahissar et al. 1992; Eggermont 1992; Eggermont 1994; decharms and Merzenich 1996; Swadlow et al. 1998; Roy and Alloway 1999; Steinmetz et al. 2000).

62 50 Part of the controversy surrounding the temporal binding hypothesis concerns whether oscillations are necessary for mediating long-range cortical synchronization (Gray 1999; Shadlen and Movshon 1999). Support for this view comes from studies indicating that gamma frequency oscillations (20-80 Hz) are prevalent in stimulusinduced cortical responses that are synchronized across distances greater than 2 mm (Eckhorn et al. 1988; Gray and Singer 1989; Konig et al. 1995; Murthy and Fetz 1996; Murthy and Fetz 1996). Other studies, however, report that only a small fraction (< 20%) of cortical spike trains contain oscillatory activity in the gamma frequency range (Tovee and Rolls 1992; Young et al. 1992; Bair et al. 1994; Nowak et al. 1995; Gray and Viana Di Prisco 1997). To some extent, discrepancies in the incidence of oscillatory activity may reflect the possibility that stimulus-induced oscillations are transitory and, therefore, more difficult to detect when responses are summed across several blocks of trials (Ghose and Freeman 1992; Livingstone 1996; Murthy and Fetz 1996; Gray and Viana Di Prisco 1997). Nonetheless, the low incidence of cortical oscillations in many studies has led some to argue that long-range correlations are rare and contribute very little to perception (Shadlen and Movshon 1999). Furthermore, it has been argued that even if cortical oscillations were prevalent, the number of independent neuronal assemblies that could be linked by oscillatory activity is limited. Synchronous activity is sometimes portrayed by paired events transpiring over intervals lasting ms or more (Gray and Singer 1989; Engel et al. 1991; Murthy and Fetz 1996; Brecht et al. 1998), and the duration of these intervals would interfere with the possibility of maintaining multiple phase relationships simultaneously. According to this line of reasoning, the temporal

63 51 binding hypothesis appears implausible unless long-range cortical synchronization can occur within extremely short time intervals lasting no more than 4 ms (Shadlen and Movshon 1999). To determine if neuronal oscillations are essential for long-range cortical synchronization in the somatosensory system, we analyzed the temporal structure of stimulus-induced neuronal responses recorded simultaneously in the forepaw representations of cortical areas SI and SII. These regions are located in different gyri and are separated by at least 10 mm (Alloway and Burton 1985). In contrast to studies that assessed the incidence and strength of oscillations independent of synchronization, our analysis was limited to neuronal pairs that showed significant levels of synchronization. Our results indicate that long-range cortical synchronization may occur within narrow time periods without concomitant neuronal oscillations in the gamma frequency range Methods Surgery Experiments on four adult cats followed NIH guidelines for the use and care of laboratory animals. Most procedures are briefly described here because they have been described previously (Johnson and Alloway 1996; Roy and Alloway 1999). Sterile techniques were used to implant a stainless steel recording chamber onto the cranium overlying SI and SII cortex. A stainless steel bolt was attached to the occipital ridge to immobilize the animal s head during recording sessions. Extracellular recordings from SI and SII were performed two times per week for 2-4 months. During recording

64 52 sessions the animals were ventilated through an endotracheal tube with a 2:1 gaseous mixture of nitrous oxide and oxygen containing % isoflurane. Heart rate and end-tidal CO 2 were monitored continuously, and body temperature was maintained at 37 C by thermostatically controlled heating pads. This preparation was similar to the anesthetized preparations used to characterize neuronal oscillations in the visual system (Gray and Singer 1989; Engel et al. 1990; Ghose and Freeman 1992; Gray and Viana Di Prisco 1997). The final experimental session was terminated by an intravenous injection of 30 mg pentobarbital sodium. The animal was transcardially perfused with 500 ml of 0.9% saline containing 20 mg lidocaine and 1,000 USP units of heparin, followed by 500 ml neutral formalin, and then 500 ml of neutral formalin in 10% sucrose. The brain was removed and placed in fixative and 30% sucrose until it sank. The cortex was blocked, frozen, and cut into 50? m coronal sections that were mounted onto chrom-alum coated slides and stained with thionin Electrophysiology Arrays of 2-8 tungsten electrodes (2-5 M? ; Frederick Haer) with ? m separation between adjacent electrodes were used to record multiple neurons in SI and SII cortex simultaneously. After placing one electrode array in the forelimb representation of SII cortex at a 50? angle to the parasagittal plane (Alloway and Burton 1985), the second electrode array was advanced into the SI forelimb representation at a 25? angle (Felleman et al. 1983). In both regions, electrodes were advanced until neurons

65 53 were encountered that responded to airjet stimulation. Electrode recording depths were consistent with neurons in layers III and IV. Receptive field (RF) boundaries of the recorded neurons were determined by stimulating the hairy skin with an airjet while monitoring each channel over an acoustic speaker. Extracellular waveforms were digitized, timestamped, and stored for off-line analysis (DataWave Technologies, Broomfield, CO). The digitized waveforms were sorted on the basis of multiple parameters (width, amplitude, time of maximum and minimum potentials, etc.), and were used to construct peristimulus timed histograms (PSTHs), cross-correlation histograms (CCGs), and auto-correlation histograms (ACGs) Cutaneous Stimulation We recorded only neurons that displayed cutaneous responses to moving the fine hairs on the distal forelimb. Neurons that responded to stimulating the glabrous skin, the claws, or responded to intense stimuli such as tapping, kneading, or pinching the skin were never recorded. While searching for airjet sensitive neurons in SI and SII, we used a fine brush or a hand-held airjet to stimulate the hairy skin. Previous work has shown that moving airjets consistently activate mechanoreceptors of the hairy skin without producing the lateral distortions caused by dragging a probe across the skin (Ray et al. 1985). We previously showed that somatosensory cortical neurons respond better to moving airjets than to stationary airjets (Roy and Alloway 1999) and, therefore, we used computer-controlled moving airjets to activate cortical neurons in SI and SII. A modified Grass polygraph pen module, in

66 54 which the ink pen was replaced with an airjet tube, was used to deliver moving airjets to the hairy skin. Airflow through the tube was controlled by an electronic valve that was gated by the data acquisition system (DataWave Technologies). Air pressure during stimulation was held constant to 20 psi by a needle valve in series with a pressure gauge. The motion of the airjet tube was controlled by a 1 Hz sine wave output from a function generator that lasted 3 seconds so that the skin was stimulated three times in each direction. The trajectory of the airjet extended across the entire length of the combined RFs, typically a distance of 3-7 cm, and this corresponded to a velocity range of 6-14 cm/sec. A portion of the airjet trajectory sometimes stimulated the glabrous skin, but in most instances the airjet was located so that its entire trajectory was over hairy skin. Airjet stimuli were presented in blocks of trials. Each trial consisted of a prestimulus period (2 s), a stimulus period with the moving airjet (3 s), and then a poststimulus period (3 s). Intervals between trials were approximately 2 seconds in duration Cross-correlation analysis Cross-correlation analysis was performed on neuron pairs in which both neurons discharged at least 12 times per stimulation period for an average rate of 4 discharges per second. Although this rate is below the gamma frequency range (20-80 Hz), a neuron discharging at least 12 times in a 3 second period can easily display gamma range oscillations if the discharges occur at interspike intervals of 50 ms or less. Raw CCGs were constructed to display changes in target neuron activity as a function of reference neuron discharges occurring at time zero (Perkel et al. 1967).

67 55 Stimulus coordination effects were removed by subtracting a linear shift predictor from the raw CCG to produce a neural CCG (Alloway et al. 1994; Johnson and Alloway 1996; Roy and Alloway 1999). The shift predictor was used to calculate 99% confidence limits, and peaks in the neural CCG that exceeded the 99% confidence limits were regarded as statistically significant (Aertsen et al. 1989; Gochin et al. 1989). Although neural CCGs were used initially to establish the statistical significance of synchronized activity, the rate and strength of neuronal synchronization were measured from the raw CCGs because these represent all the events available for sensory processing. The proportion of correlated activity among pairs of neurons can be estimated by the correlation coefficient,?(?). The formula for calculating the cross-correlation coefficient was similar to the formula used by Eggermont (1992):? (? )?? CE? 2 2?? N? (( N ) / T )???*? N? (( N ) / T )??? T T R R where? represents the bin size over which the coefficient is evaluated (3 ms), CE is the number of coincident events in the highest 3 ms period of the raw CCG peak, T is the time interval over which the CCG was calculated, and N T and N R represent the total number of discharges from the target and reference neurons during T. In contrast to studies in which CE represents the number of coincident events in the neural CCG (Eggermont 1992), we measured CE in the raw CCG. This modification meant that our correlation coefficients represent the proportion of all activity that was correlated, not just the portion exceeding the expectation density.

68 56 The correlation coefficient indicates the proportion of activity that is correlated, but does not indicate the overall rate of coincident events. Therefore, we also calculated the rate of coincident events in SI and SII to provide another measure of synchronization strength during spontaneous and stimulus-induced activity. Synchronization rate was determined by counting the number of coincident events in the tallest 3 ms period of the raw CCG peaks generated from spontaneous or stimulus-induced activity. These sums were then divided by the amount of time over which spontaneous or stimulus-induced responses were recorded (Roy and Alloway 1999). The spontaneous rates were always based on the 2 second pre-stimulus periods; the stimulus-induced rates were based on the portion of the stimulation cycle in which both neurons responded simultaneously to cutaneous stimulation Analysis of Neuronal Oscillations Neurons displaying synchronized activity were carefully examined for the presence of gamma oscillations in their discharge sequence. As in previous reports on stimulus-induced oscillations, we estimated an empirical power spectrum by computing the fast Fourier transform (FFT) of the neuronal ACGs derived from the stimulus-induced responses (Ghose and Freeman 1992; Gray and Viana Di Prisco 1997). All ACGs were constructed with a time lag of 0 to 256 ms and a binwidth resolution of 1 ms. A power spectrum derived from an ACG with these parameters has a maximum frequency of 500 Hz and a bin resolution of 3.9 Hz.

69 57 When a FFT is computed on a Gaussian white noise sequence, the resulting power spectra should be relatively flat and show random variations around a mean value. In some cases, as shown in Figure 3.1, the power spectrum that was derived from the raw ACG contained a broad DC-low frequency component that resembled non-white noise. These low frequency components were wider than the low frequency signals appearing in the power spectra of earlier studies on the visual system (Ghose and Freeman 1992; Gray and Viana Di Prisco 1997), possibly because our cortical responses were correlated with low frequency hair movements evoked by the airjet stimulus. In any event, peaks in the Hz range often appeared just above the confidence limits (see below) of the power spectrum, but it was unclear whether these peaks represented significant amounts of oscillatory power or were insignificant because they were superimposed upon a broader low frequency component. Therefore, we used well-established techniques in signal processing to adjust the ACG so that we could analyze its gamma frequency components independent of effects caused by low frequency spectral leakage (Oppenheim and Schafer 1989). First, the impulse-like peak of the ACG near t = 0 (up to 6 ms) was removed because this narrow peak contains broadband spectral energy, including low frequency components. Removing time intervals representing the first 6 ms of the ACG, however, should not affect the power of gamma range signals because these intervals represent high frequency neuronal bursting (> 165 Hz). Second, the mean bin height of the original ACG was calculated and subtracted (Oppenheim and Schafer 1989). In addition, any linear trend in the ACG was removed by subtracting a least squares regression line (Oppenheim and Schafer 1989). These adjustments produced a truncated

70 Raw ACG (SI - A73) Truncated ACG ms 7.5 Power Spectrum of Raw ACG (0-256 ms) Candidate Oscillation at 34.6 Hz (S/N = 3.04) Hz Power Spectrum of Truncated ACG (6-256 ms) Normalized Power Spectrum Hz 0.5 S/N = 2.68, p > Hz Figure 3.1 Method for detecting neuronal oscillations. Top panel: Raw and truncated ACGs for a spike train recorded in SI (A73). With the exception of the impulse peak at time zero, the same rhythmic patterns are present in both the raw and truncated ACGs. Second panel: Power spectrum of the raw ACG contains a large DC component that gradually merges into the gamma frequency range (20-80 Hz). Although power fluctuates randomly around the noise level (dashed line) at frequencies greater than 65 Hz (arrowhead), a peak at 34.6 Hz exceeded the confidence limits (dotted line) and had a signal to noise ratio of This small peak may have exceeded the confidence limits because it was superimposed upon the trailing edge of an elevated low frequency trend. Third panel: Power spectrum of the truncated ACG contained a much smaller DC component, but the low frequency trend was still present. Bottom panel: The normalized power spectrum contained fluctuations around the noise level at frequencies as low as 42 Hz (arrowhead). After normalization, the small peak at 34.6 Hz failed to exceed the confidence limits and its signal-to-noise ratio decreased to Statistical analysis indicated that the distribution of values in the gamma frequency range of the normalized power spectrum was not significantly different from a Gaussian white noise distribution having the same mean value (KS test, p > 0.01).

71 59 ACG that was considerably flatter than the original ACG but with the same temporal rhythms that were apparent in the original pattern (see first panel in Figure 3.1). Power spectra estimates were computed from the resulting truncated ACG with a Hamming window to further reduce spectral leakage (Oppenheim and Schafer 1989). Despite these adjustments to the ACGs, a few of the resulting power spectra still contained broad, low frequency components that suggest the presence of colored or nonwhite noise (see third panel in Figure 3.1). Many types of noise, including Gaussian white noise, can be characterized using an ideal power law model of the form: Power = cf*frequency -? where cf determines the overall power and? determines spectral balance (Veitch and Abry 1999). Thus, in cases where the power spectra of the truncated ACGs contained elevations in the low frequency range, we estimated the cf and? parameters from the truncated ACGs. Most of these ACGs exhibited good fits to this power law model (chi-square values greater than 0.1), and had??values greater than zero, which suggests that their power spectra did not reflect Gaussian white noise. The presence of non-gaussian noise in a power spectrum precludes the use of parametric measures such as variance to evaluate statistical significance. However, non-gaussian spectra can be converted to Gaussian spectra by a standard normalization technique (Brockwell and Davis 1991). Specifically, the non-white power spectra were divided by the theoretical power law spectra so that the resulting normalized spectra, which varied around a value of one, resembled Gaussian white noise sequences (Brockwell and Davis 1991). These normalized power spectra were further normalized by defining the largest peak as having a value of one. Figure 3.1 shows how these adjustments affect the

72 60 resulting power spectrum estimate and the classification of the neuronal spike train as oscillatory or non-oscillatory (compare second, third, and bottom panels of Figure 3.1). After obtaining power spectra that resembled white noise, those that contained peaks in the gamma frequency range (20-80 Hz) were subjected to both parametric and non-parametric statistical analysis to determine if the peaks were likely to be significant. Previous reports assumed that the Hz region of the power spectrum reflects Gaussian noise in the spike train (Ghose and Freeman 1992; Gray and Viana Di Prisco 1997). As in these previous reports, we identified potential oscillations by constructing confidence limits that were equal to the mean plus 3 standard deviations of the values appearing in the Hz portion of the power spectrum. These confidence limits were displayed on the power spectra of the truncated ACGs and, if necessary, on the normalized power spectra. If peaks in the Hz range exceeded the confidence limits, then the non-parametric Kolmogorov-Smirnov (KS) test was used to evaluate the statistical significance of candidate oscillations by comparing them to a uniformly flat, ideal power spectrum having the same mean value as the whitened spectra (Brockwell and Davis 1991). Peaks that were superimposed upon distributions that differed significantly from the ideal power spectrum (p < 0.01, KS test) were classified as oscillatory. We also analyzed CCGs to determine if the relative timing of activity across SI and SII was characterized by oscillatory patterns. The procedure for analyzing oscillations in the CCGs was the same as that used to analyze the ACGs except the CCGs

73 61 were not truncated, and the analysis was conducted over all events in the CCG extending from 256 ms to +256 ms Results Airjet-sensitive responses containing at least 12 discharges per stimulus were recorded from 228 neurons in SI and 314 neurons in SII. From this sample, the number of neuron pairs recorded simultaneously in SI and SII totaled 621. Using 99% confidence limits, cross-correlation analysis revealed significant levels of stimulusinduced synchronization in 80 SI-SII neuron pairs or 13% of the total sample. Some neurons participated in several synchronized pairs, and the constituent neurons of all 80 synchronized pairs included 67 neurons in SI and 69 neurons in SII. Pairs of neurons displayed synchronized activity only if their RFs were highly similar. In most cases of SI-SII synchronization, both neurons shared at least one half of their RFs. Typically, the RF of the SI neuron was located on the ventral surface of one or two digits, whereas the RF of the SII neuron was larger and completely surrounded the RF of the SI neuron. The only exception to this occurred in experiments in which the SI neurons had larger RFs because they were located in the wrist representation of SI cortex (see Felleman et al. 1983) Synchronization of Single Unit Activity in SI and SII An example of long-range synchronization in SI and SII cortex is illustrated in Figure 3.2. As indicated by the RF drawings and PSTHs, neurons SI-A151 and SII-A151 had overlapping RFs on the distal forepaw and responded to airjets that traversed the

74 A B 700 SII-A Events / Bin SI-A C 600 Spontaneous Moving Airjet D SII-A s Correlated Events Events / Bin SI-A Hz S/N = ms ms Hz Figure 3.2 Synchronized activity in a pair of SI and SII neurons in which one neuron contained weak oscillations in the gamma frequency range. A. Extracellular waveforms and receptive fields for a pair of simultaneously recorded neurons in experiment A151. The waveforms represent the average shape of all discharges recorded during spontaneous and stimulus-induced activity as shown in Panel B. Trace duration, 1.5 ms; scale bars, 50 µv for SII, 100 µv for SI. Drawing of the distal forepaw illustrates the trajectory of the 1 Hz moving airjet as it crosses the overlapping receptive fields of each neuron. B. PSTHs illustrating spontaneous and stimulus-induced responses across 300 trials. The arrows below the bottom PSTH illustrate the back-and-forth motion of the airjet as it repeats three consecutive 1 Hz cycles. Binwidths, 25ms. C. Raw (top) and shift-corrected (bottom) CCGs displaying correlated discharges recorded during spontaneous and stimulus-induced activity. Dotted lines in the shift-corrected CCGs indicate 99% confidence limits. Binwidths, 1.0 ms. D. Raw ACGs (left) constructed from stimulus-induced responses during the moving airjet. Although the ACGs were constructed over lag intervals of 256 ms, only the first 128 ms are displayed to facilitate detection of oscillations occurring in the gamma frequency range (20-80 Hz). Neuron SI-A151 contains a small peak at ms which corresponds to a frequency of Hz. Analysis of the power spectra (right) derived from the truncated ACGs revealed a significant oscillatory component at Hz (KS test, p < 0.007) for neuron SI-A151. ACG binwidths, 1.0 ms; power spectra binwidths, 3.9 Hz. Dashed and dotted lines in the power spectra indicate mean noise and confidence limits (i.e. mean noise plus 3 standard deviations), respectively.

75 63 hairy skin of their RFs. In this case, the RF for the SI neuron extended across the ventral surface of digit 3, and was completely encompassed by the RF of the SII neuron, which included the ventral surface of digits 2, 3, and 4. Consistent with this fact, comparison of the PSTHs indicates that the response of the SI neuron was restricted to a smaller portion of the stimulus cycle than the response of the SII neuron (i.e. 825 ms of the entire stimulation period). Furthermore, because we used a large, periodic airjet to stimulate relatively small RFs, both neurons displayed low frequency components in their response pattern as indicated by prominent peaks appearing every ms in the PSTHs. The amplitudes of the PSTH peaks show that the maximum response rate of both neurons was approximately spikes per second, and this demonstrates that much of the stimulusinduced activity occurred within the gamma frequency range (20-80 Hz). Cross-correlation analysis revealed that both neurons exhibited substantial amounts of synchronized activity during airjet stimulation but not during pre-stimulus periods (Figure 3.1C). Thus, the neural CCG constructed from the stimulus-induced responses contained a tall peak at time zero that was approximately 12 ms in duration at its base and 4 ms in duration at half its peak height. Based on the number of coincident events occurring within the peak of the raw CCG and the duration of simultaneous responses in both neurons, the synchronization rate was 5.4 coincident events per second. The correlation coefficient, which provides a rough estimate of the portion of stimulusinduced activity that was synchronized, was 0.12 (or 12%) for this pair of neurons. These indices of synchronization strength indicate that this pair of neurons was among the most highly synchronized pair of neurons that we recorded in our sample.

76 64 Analysis of the ACGs and corresponding power spectra indicated that one of the neurons (SI-A151) contained relatively weak oscillations in the gamma frequency range and that both neurons contained stronger oscillations in the low frequency range (Figure 3.1D). Thus, a small peak in the ACG for the SI neuron was apparent at a time lag of ms. Computation of the power spectrum from the truncated ACG revealed a small peak (ranging from Hz) with a signal-to-noise ratio (3.2) that barely exceeded the confidence limits calculated from randomly-fluctuating values between 250 and 500 Hz. Because this power spectrum consisted primarily of white noise (note the presence of valleys extending to the mean noise level at frequencies below 46 Hz), computation of the normalized power spectrum was not necessary. Statistical analysis of the distribution of values in the gamma frequency range confirmed that these values were significantly different from an idealized power spectrum (p < , KS test). The ACG and power spectrum for the other neuron (SII-A151) did not contain oscillations in the gamma frequency range, but the power spectra for both neurons contained obvious peaks at 4-8 Hz. In addition to representing possible spectral leakage from the DC component, these low frequency peaks are consistent with the fact that the PSTHs for both neurons contained rhythmic patterns that correspond to the periodicity of the airjet stimulus. Figure 3.3 illustrates synchronized responses for a representative pair of SI and SII neurons (A-41) that did not oscillate in the gamma frequency range. Both of these neurons had overlapping RFs and responded vigorously as the airjet moved across their RFs in either direction. As indicated by the PSTHs, the maximum response rate of both neurons was approximately 30 discharges per second, which is clearly within the gamma

77 65 A B 40 SII-A41 20 Spikes / Sec 0 40 SI-A41 20 C 200 Spontaneous Moving Airjet D SII-A s Correlated Events Events / Bin SI-A ms ms Hz Figure 3.3 Synchronized activity in a pair of SI and SII neurons (A-41) that did not oscillate in the gamma frequency range (20-80 Hz). Neuronal waveforms, receptive fields, and temporal patterns of spontaneous and stimulus-induced activity are illustrated as in Figure 3.2. In panel A, the waveform traces represent a duration of 1.2 ms, and the scale bar represents 100 µv for both waveforms. The clear difference in the waveform patterns indicates that the temporal precision of synchronized activity (see Panel C) was not due to electrical artifacts.

78 66 frequency range. Their PSTHs also displayed rhythmic patterns that correspond to the low frequency periodicity of the cutaneous stimulus. Consistent with the fact that the RFs and response properties of these neurons were similar, the raw and neural CCGs revealed an extremely narrow peak of synchronized activity at time zero. Thus, as indicated by the neural CCG, the temporal duration of the peak halfwidth was only 1 ms while the width of the peak at its base was only 3 ms in duration. We examined the waveforms of the neuronal discharges in SI and SII (see Figure 3.3A), and observed that the amplitude and shape of the waveforms were different; this effectively ruled out the possibility that the narrow peaks in the raw and neural CCGs were due to electrical artifacts. The mean synchronization rate (0.74 coincident events per second over the entire stimulation period) and correlation coefficient (p(?) = 0.053) for this pair of neurons placed it in the top 34 percent for synchronization strength when compared to all synchronized SI-SII neuron pairs. Examination of the ACGs and power spectra failed to indicate the presence of gamma range oscillations in the spike trains of SI-A41 or SII-A41. The ACGs for both neurons contained substantial levels of activity at lag times ( ms) that correspond to the gamma frequency range (20-80 Hz), but the fluctuations in the temporal structure of the ACGs were sporadic and did not appear to represent non-random periodicities. This belief was corroborated by the fact that the power spectra of the truncated ACGs did not contain any peaks in the Hz range that exceeded the confidence limits. The major peak seen in both power spectra, especially for neuron SII-A41, was at a frequency of 4-8 Hz.

79 Synchronization of Multiunit Activity in SI and SII Many studies that reported observing neuronal synchronization in the visual system were based on multiunit responses (Eckhorn et al. 1988; Gray and Singer 1989; Engel et al. 1990; Gray et al. 1992), and this is not surprising because synchronization is easier to detect in multiunit responses than among pairs of individual neurons (decharms and Merzenich 1996; Bedenbaugh and Gerstein 1997; Roy and Alloway 1999). Therefore, we also evaluated multiunit responses in SI and SII to determine if oscillations were more likely to be detected in the synchronized activity of small populations of neurons. Multiple isolated waveforms were recorded from a total of 99 electrodes in SI and 146 electrodes in SII; in all of these cases at least two neurons recorded by each electrode discharged at least 12 times per stimulus. Although a single electrode sometimes recorded 5 distinct waveforms, on average we recorded only 2.4 neurons per electrode. From this sample of multiunit responses, the number of multiunit pairs recorded simultaneously in SI and SII totaled 118. Using 99% confidence limits, crosscorrelation analysis revealed significant levels of stimulus-induced synchronization in 29 pairs or 25% of the total sample. These 29 synchronized multiunit pairs were based on 24 and 28 constituent multiunit responses in SI and SII, respectively. Examples of stimulus-induced multiunit responses in SI and SII are illustrated in Figure 3.4. Both of the SI and SII electrodes in this example recorded discharges from a pair of cortical neurons. In this particular case, the RFs of the two SI neurons (ulnar paw and wrist) completely encompassed the RFs of the SII neurons, which were restricted to digit 5. Consistent with these RF differences, the PSTHs indicate that the SII responses

80 68 A B 100 SII-A Spikes / Sec SI-A C 800 Spontaneous Moving Airjet D s SII-A Correlated Events Events / Bin SI-A ms ms Hz Figure 3.4 Synchronized multiunit responses in SI and SII (A-130) without concomitant oscillations. Temporal patterns of spontaneous and stimulus-induced multiunit responses are illustrated as in Figures 3.2 and 3.3. As indicated by the ACGs and power spectra in panel D, stimulus-induced responses in both SI and SII were devoid of oscillations in the gamma frequency range (20-80 Hz).

81 69 were restricted to a smaller portion of the stimulus cycle than the SI responses. Nonetheless, both PSTHs contained peaks at regular intervals ( ms) that corresponded to the periodicity of the moving airjet. In addition, the height of these regular peaks indicates that stimulus-induced responsiveness reached a maximum rate that easily achieved the gamma frequency range for both sets of neuronal responses (up to 140 and 80 spikes/s for SI-A130 and SII-A130, respectively). Cross-correlation analysis of these responses revealed substantial amounts of synchronized activity during airjet stimulation, but not during spontaneous activity (see Figure 3.4C). The stimulus-induced neural CCG contained a significant peak that was maximal 0-3 ms after time zero and had a peak half-width of 4 ms. Analysis of the same 3 ms period in the corresponding peak of the raw CCG revealed a correlation coefficient of 0.25 and a mean synchronization rate of 10.9 coincident events/s over the entire stimulus period. These stimulus-induced multiunit responses were not associated with oscillations in the gamma frequency range (see Figure 3.4D). Thus, the multiunit ACGs did not contain any prominent periodicities in the intervals extending up to 128 ms. Consistent with the smooth shape of the multiunit ACGs, the power spectra for the truncated ACGs did not contain any peaks between 20 and 80 Hz that exceeded the confidence limits. In fact, the only peaks in the power spectra that exceeded the confidence limits represented frequencies around 4 Hz.

82 Distribution of Single and Multiple Neuron Synchronization The strength of synchronized responses in SI and SII varied tremendously across pairs of recording sites. As indicated by cumulative distributions in Figure 3.5, both the single and multiple neuron responses displayed a hundred-fold difference in the rate of coincident events when the weakest and strongest neuronal pairs were compared. Similarly, the cumulative distribution of correlation coefficients indicated that the proportion of synchronized activity varies by a factor of ten when comparing the strongest and weakest cases of correlated activity. The margin of error for placing electrodes in corresponding parts of SI and SII is fairly small as indicated by data showing that focal cutaneous stimulation causes synchronization among local populations of cortex that extend no more than 500 or 600 microns in diameter (Metherate and Dykes 1985; Roy and Alloway 1999). Thus, weakly synchronized responses seem likely to represent pairs of SI and SII neurons whose locations are on the perimeter of regions Cumulative Proportion of Synchronized Pairs (%) Multiple Neuron Pairs Single Neuron Pairs Spontaneous Stimulus-Induced Synchronization Rate (events/s) Correlation Coefficient Figure 3.5 Cumulative distributions illustrating the strength of spontaneous and stimulus-induced synchronization across SI and SII. Data are based on 80 pairs of single neurons and 29 pairs of multiple neurons in which significant peaks appeared in the neural CCGs of the stimulus-induced responses. Maximum synchronization strength for each group is indicated by filled and unfilled triangles appearing along the bottom axis.

83 71 showing the strongest interactions. Hence, the highest values for the correlation coefficients and synchronization rates probably provide the best measure of synchronization that is typically achieved at optimal pairs of recording sites in SI and SII. Analysis of the temporal structure of SI and SII coordination indicates that stimulus-induced synchronization was associated with a relatively high degree of temporal precision. As shown in Figure 3.6, almost 60% of the single neuron pairs had peak halfwidths that were 5 ms or less, and 90% had halfwidths that were 10 ms or less. Furthermore, the peaks in the neural CCGs were distributed close to time zero. Thus, for both single and multiple neuron pairs, the tallest bin of the neural CCG peaks was located within 5 ms of time zero for 50% of the cases, and the most common time lag was only 1 or 2 ms (n = 42). These data indicate that stimulus-induced synchronization in SI and SII is comprised largely of discharges that occur at about the same time. Figure 3.6 Precision of temporal synchronization for single unit and multiunit stimulus-induced responses in SI and SII. Top graph: distribution of peak halfwidths for pairs of synchronized single and multiple neuron responses. A peak halfwidth is defined as the temporal width of a peak in the neural CCG at half its height. Bottom graph: distribution of neural CCG peak times with respect to time zero. Number of Synchronized Pairs Peak Halfwidth (ms) Single Neuron Pairs Multiple Neuron Pairs Peak Time (ms)

84 72 Number of Oscillations 10 5 Single Neurons Multiple Neurons Frequency (Hz) Figure 3.7 Distribution of neuronal oscillations according to frequency. Stimulus-induced neuronal responses that exhibited significant levels of oscillations in the gamma frequency range are represented according to their tallest peak in the power spectrum derived from the ACG. Binwidths appear as 4 Hz increments to correspond with the frequency resolution of the power spectra Incidence of Neuronal Oscillations in the ACGs Analysis of the power spectra derived from the single and multiple neuron ACGs indicated that only a small fraction were associated with significant power levels in the gamma frequency range. Among 136 neurons in SI and SII that formed 80 synchronized pairs during airjet stimulation, only 13% (n = 18) of these neurons contained oscillations between 20 and 80 Hz. Among the 52 multiunit responses that formed 29 synchronized pairs, only 15% (n = 8) were characterized by oscillations. Detectable oscillations spanned the entire spectrum of the gamma frequency range but, as Figure 3.7 indicates, the majority of these oscillations represented frequencies between 20 and 32 Hz. All of the oscillations appeared to be weak and, as shown in Figure 3.8, the power of Signal to Noise Ratio SI neurons SII neurons Figure 3.8 Strength of neuronal oscillations. Each bar represents the mean signal-to-noise ratio for significant oscillations detected in the gamma frequency range of the power spectrum. Noise was calculated as the average power level in the Hz range. Error bars indicate SEM. 0.0 Single Neurons Multiple Neurons

85 73 Table 3.1 Incidence of oscillations during SI-SII synchronization Single Unit Pairs Multiunit Pairs No Oscillations SI only 11 4 SII only 7 4 Both different Hz 0 0 Both same Hz 1 0 Total oscillations in the gamma frequency range was typically only 2-3 times greater than the mean amount of power occurring in the high frequency ( Hz) portion of the power spectrum. Several facts suggest that stimulus-induced synchronization in SI and SII does not depend on oscillations in the gamma frequency range. As indicated by Table 3.1, over 76% (61/80) of the synchronized neuron pairs were devoid of oscillations and 22% (18/80) of the neuron pairs displayed oscillations in only one neuron. The low incidence of oscillations among synchronized responses is underscored by the fact that in only one pair of synchronized neurons did both cells oscillate at the same frequency. Essentially similar results were obtained among the pairs of synchronized multiunit responses (see Table 3.1). Furthermore, and contrary to our expectations, the presence of oscillations did not increase the strength of synchronized activity in SI and SII. As summarized in Figure 3.9, mean synchronization rates and correlation coefficients were lower when oscillations were present in one or both of the constituent responses. Statistical analysis failed to reveal any difference in the discharge rates of oscillatory and non-oscillatory neurons (Table 3.2). Thus, the lower

86 Synchronization Rate (coincident events/s) Correlation Coefficient No Oscillations SI or SII Oscillating Both - Same Hz Single Unit Pairs Multiunit Pairs 74 Figure 3.9 Strength of synchronized activity in SI and SII as a function of the presence or absence of oscillations in the constituent neuronal responses. Top graph: mean synchronization rate for single unit or multiunit pairs in which oscillations were detected in neither, one, or both of the neuronal responses as indicated by the legend. Bottom graph: mean magnitude of correlation coefficients for the same groups shown in the top graph. Error bars indicate SEM. synchronization strength apparent among oscillating responses was not due to differences in the mean rate of activity. We also searched for high frequency oscillations in individual stimulus trials because oscillations may occur momentarily and go undetected in the summed ACGs, especially if the predominant oscillatory frequency varies across trials (Ghose and Freeman 1992; Murthy and Fetz 1996; Gray and Viana Di Prisco 1997). Oscillations within a single trial cannot be detected in spike trains that contain low rates of activity and, therefore, our analysis was limited to the most responsive multiunit recording Table 3.2. Stimulus-induced activity of synchronized neurons in SI and SII SI SII Oscillating neurons 20.8 ± ± 5.3 Non-oscillating neurons 18.9 ± ± 1.9 Numbers indicate discharges/s (mean ± SEM)

87 75 experiments, such as the one depicted in Figure 3.4 (experiment A130). A trial-by-trial analysis revealed that the vast majority of responses in experiment A130 did not contain oscillations in the gamma frequency range. As indicated by the ACG and power spectrum obtained for each stimulus trial, responses for both SI-A130 and SII-A130 were devoid of gamma frequency oscillations in 59 of the 100 trials. Rhythmic activity between 20 and 80 Hz was detected by our statistical analysis in 41 trials, and most of these oscillations were limited to one electrode in SI (n = 8) or SII (n = 29). Only four trials were characterized by oscillations on both electrodes, but often at different frequencies. Hence, among 200 ACGs analyzed in this experiment, only 22.5% (45/200) contained significant levels of gamma range oscillations. As shown in Figure 3.10, which illustrates those ACGs that contained the strongest gamma oscillations detected in experiment A130, most of the Hz peaks in the power spectra had very low signalto-noise ratios. In fact, visual inspection revealed clear instances of oscillations in only three ACGs (SI on trials 39 and 49, and SII on trial 58, Figure 3.10). To determine the relative contribution of oscillations in synchronizing the SI and SII responses in experiment A130, we performed cross-correlation analysis separately on those trials classified as containing no oscillations, oscillations in only one cortical area, or oscillations in both SI and SII. The results of this analysis are illustrated in Figure When the individual CCGs shown in Figure 3.11 are summed together, the resulting CCG matches the neural CCG shown in Figure 3.4 (bottom right of panel C). A comparison of the CCGs shown in Figure 3.11 indicates that each type of trial contributed some synchronized activity (events at 0-3 ms after time zero), but most of the

88 SI Responses SII Responses Hz S/N = Hz S/N = Trial Hz S/N = Trial Hz S/N = Trial Hz S/N = Trial Hz S/N = Trial Hz S/N = Trial Time (ms) Frequency (Hz) Time (ms) Frequency (Hz) Figure 3.10 Trial-by-trial analysis of oscillatory responses in experiment A130. These trials illustrate the strongest oscillations (in terms of signal-to-noise ratio) that were detected among 100 trials whose summed multiunit responses are illustrated in Figure 3.4. The ACGs and power spectra represent the temporal structure of activity recorded simultaneously from SI (left) and SII (right) during specific trials as indicated. Arrows indicate peaks in the power spectra that represent significant oscillations in the gamma frequency range. Solid and dashed horizontal lines represent mean noise (from Hz) or mean noise plus three standard deviations, respectively. Visual inspection of all 200 ACGs revealed gamma range oscillations in SI on trials 39 and 49, and in SII on trial 58.

89 77 synchronized events were produced in trials that contained no oscillations or else contained oscillations in SII alone. Given the scaling differences of these CCGs, it appears that the number of synchronized events contributed by each category was proportional to the number of trials Incidence of Oscillations in the CCGs We also analyzed the power spectra derived from single and multiple neuron CCGs to determine if there were periodicities in the relative timing of discharges in SI and SII. Among 80 raw CCGs based on single neuron responses, only two contained significant levels of oscillations in the Hz range (experiment A64 at 25 Hz, experiment A101 at 33 Hz). Similarly, among 29 raw CCGs based on multiple neuron responses, only one contained a significant level of oscillatory activity (experiment A96 at 58 Hz). 100 No Oscillations 40 SI Only 80 SII Only 30 Dual Oscillations Correlated Events Time (ms) Figure 3.11 Neural CCGs illustrating the degree of synchronization in experiment A130 for those trials classified as containing no oscillations (n = 59), oscillations in SI only (n = 8), in SII only (n = 29), or in both cortical areas (n = 4). Because of differences in the number of trials, the CCGs are scaled so that the 99% confidence limits span approximately the same distance in each histogram.

90 Discussion This study revealed two important findings regarding the coordination of stimulus-induced activity in separate cortical areas. First, we found that tactile stimulation evokes synchronous responses in corresponding somatotopic representations of SI and SII cortex. These responses are extremely precise, usually occurring within intervals of 5 ms or less. Second, we frequently observed significant levels of long-range synchronization between SI and SII without the presence of oscillations in the constituent neurons. In instances where gamma range oscillations were detected, periodic activity was usually present in one neuron of a synchronized pair. We rarely observed simultaneous oscillations in both SI and SII at the same frequency. These findings suggest that separate populations of cortical neurons can be bound together by a sensory stimulus to form functional assemblies without being constrained by the phase relationships defined by specific oscillatory frequencies Variable and Transient Incidence of Synchronized Oscillations Neuronal oscillations have received much attention as a potential mechanism for grouping distributed populations of cortical neurons, but synchronized oscillations are often transient and may not be detected if responses are summed across several blocks of trials. In the visual system, for example, experiments on both cats and monkeys have shown that stimulus-induced neuronal oscillations may occur at different frequencies on different trials(ghose and Freeman 1992; Livingstone 1996; Gray and Viana Di Prisco 1997). Furthermore, gamma range oscillations in the sensorimotor cortex of monkeys are

91 79 correlated with certain manipulative behaviors, but these oscillations occur episodically throughout the behavioral task (Murthy and Fetz 1996; Murthy and Fetz 1996). These findings raise the concern that we did not detect neuronal oscillations because our ACGs were generated from spike trains acquired over multiple trials, and this approach may have concealed oscillations that occurred at different frequencies on different trials. Several facts, however, argue against this possibility. First, even though oscillations may appear transiently, the overwhelming majority of oscillatory responses reported in the visual system were based on cumulative ACGs generated from multiple trial spike trains. Our cumulative ACGs and their derived power spectra never displayed oscillations as prominent as those seen in visual responses that were acquired over multiple trials (Gray et al. 1989; Engel et al. 1991; Gray and Viana Di Prisco 1997). Second, even when we searched for periodic activity during single trials, most of the single trial ACGs lacked oscillations in the gamma frequency range. In fact, only about 2% of the single trial ACGs contained oscillatory activity that was perceptible by visual inspection; the remaining oscillations were barely detected by rigorous statistical analysis. Finally, oscillations in the Hz range rarely appeared in SI and SII simultaneously, but were confined almost exclusively to one cortical area or the other. It is also conceivable that we did not observe prominent gamma-range oscillations because our preparation was anesthetized, but the evidence does not support this interpretation. The earliest studies reporting neuronal oscillations in the visual system involved anesthetized cats (Gray and Singer 1989; Engel et al. 1990; Ghose and Freeman 1992; Nowak et al. 1995), and this prompted the view that neuronal oscillations might

92 80 reflect an artifact of anesthesia. A direct comparison of oscillating neurons in awake and anesthetized cats, however, indicated that the incidence, frequency, and amplitude of oscillating responses were similar in both preparations (Gray and Viana Di Prisco 1997). Furthermore, with the use of rigorous statistical analysis, we detected gamma range oscillations in 15% of our neuronal sample which corresponds almost exactly with the incidence of oscillations reported in the visual cortex of anesthetized cats (see Table 1 in Gray and Viana Di Prisco, 1997) Periodic and Aperiodic Cortical Synchronization Although oscillations are thought to facilitate long-range cortical synchronization, our results indicate that oscillations are not always inherent among synchronized responses. Long-range synchronization may involve periodic or aperiodic neuronal activity depending on the connectivity of the cortical network or the type of cells that are recorded (Gray and McCormick 1996). In the visual system, some investigators observed gamma frequency oscillations in the lateral geniculate nucleus that could entrain similar oscillations in the visual cortex (Ghose and Freeman 1992). By contrast, we previously reported that stimulus-induced responses in the ventrobasal thalamus and SI cortex are precisely coordinated, but we rarely observed any neurons that oscillated in the gamma frequency range (Johnson and Alloway 1994; Johnson and Alloway 1996). Periodic or aperiodic synchronization across cortical areas might also depend on the context in which the cortical network is activated. Although some suggest that neuronal oscillations in visual cortex do not vary with different stimulus parameters (Ghose and Freeman 1992), others indicate that oscillatory amplitude is altered by

93 81 changes in stimulus velocity or luminance (Nowak et al. 1995; Gray and Viana Di Prisco 1997). Similarly, significantly fewer episodic oscillations were observed in monkey sensorimotor cortex when the animals performed repetitive wrist movements than when they performed tasks that required skill and attention (Murthy and Fetz 1996). These findings suggest that oscillatory activity is more likely if a large part of the cortical network is involved in processing the sensory attributes of a salient stimulus. In this context, it is plausible that we did not observe synchronized oscillations between SI and SII cortical areas because we did not use an optimal stimulus. Compared to the elongated, spatially extensive stimuli used to evoke synchronized oscillations in the visual system, we used a relatively focal stimulus to activate a discrete region of skin. One hypothetical function of synchronization is to link adjacent neuronal populations within a cortical area that represent co-linear stimulus attributes, such as the edge of a visual object (Singer and Gray 1995). This function, however, would not apply to a focal stimulus that activates a focal region of cortex. Another proposed function of synchronized oscillations is to link separate cortical areas that are specialized for processing different sensory features of the same stimulus. In the visual system, for example, synchronized oscillations could link separate neuronal populations that respond primarily to the color, shape, or motion of an object. A focal stimulus like a moving airjet also has cutaneous attributes, such as texture, intensity, location, and velocity of movement, which could be differentially processed by neural circuits in SI and SII cortex. Although the differential coding functions of the SI and SII cortical areas are not understood, our data demonstrate that focal airjets evoke synchronized responses in

94 82 corresponding somatotopic areas of both cortices. Whether these synchronized populations would show oscillations in response to a spatially extensive stimulus remains an empirical question. Nonetheless, our current results support the view that long-range synchronization in SI and SII is a mechanism for binding separate neural populations that represent attributes of the same stimulus Plausible Mechanisms of SI-SII Synchronization The most likely neural circuits for mediating correlated activity in cat SI and SII involve thalamocortical and corticocortical pathways. Substantial evidence suggests that common inputs from the thalamus play a major role in coordinating these cortical areas. In cats, both SI and SII receive parallel projections from neurons in overlapping parts of the ventrobasal thalamus, and 10-15% of these thalamocortical neurons send collateral projections to both SI and SII (Hand and Morrison 1970; Saporta and Kruger 1979; Spreafico et al. 1981; Bentivoglio et al. 1983; Burton and Kopf 1984). Furthermore, cross-correlation analysis demonstrates that pairs of adjacent thalamic neurons, which have similar RF and submodality properties, discharge synchronously during cutaneous stimulation (Alloway et al. 1995; Roy and Alloway 1999). In addition, stimulus-induced responses in cat SI and SII are tightly correlated with neuronal activity in the thalamus (Johnson and Alloway 1994; Johnson and Alloway 1996; Roy and Alloway 1999). Consistent with these parallel sets of thalamocortical projections, reversible inactivation of either SI or SII does not alter responsiveness in the other cortical area (Burton and Robinson 1987; Turman et al. 1992; Turman et al. 1995).

95 83 Corticocortical connections might also facilitate SI-SII synchronization. There are substantial interconnections between corresponding representations in cat SI and SII (Alloway and Burton 1985; Manzoni et al. 1990; Schwark et al. 1992), and direct electrical stimulation of SI enhances responsiveness in SII (Manzoni et al. 1979). Although SI or SII responses persist following inactivation of the other area, the timing and magnitude of the short latency responses are altered (Burton and Robinson 1987). The importance of corticocortical connections has been demonstrated in the visual system where interhemispheric connections through the corpus callosum are needed to mediate synchronization in regions that represent visual fields near the vertical meridian (Engel et al. 1991; Munk et al. 1995). The role of callosal projections in the visual system, however, might not be comparable to the role of ipsilateral connections between SI and SII because visual areas in different hemispheres do not receive common thalamic inputs. Thus, the contribution of corticocortical connections for synchronizing SI and SII might be minor when compared to the influence exerted by common projections from the ventrobasal thalamus Hypothetical Roles for Synchronization in SI and SII Synchronization among distributed populations of neurons is thought to be important for both neurotransmission and sensory perception (von der Malsburg 1994). We hypothesize that common targets of SI and SII respond preferentially to synchronized discharges in these cortical areas, just as neurons in visual or somatosensory cortex are more likely to discharge in response to synchronous activity in the thalamus (Alonso et al. 1996; Usrey et al. 2000). Thus, the sensorimotor portion of the neostriatum, which

96 84 receives convergent inputs from corresponding representations in SI and SII (Alloway et al. 2000), may depend on synchronized activity in these cortical areas in order to be activated. With respect to sensory perception, local synchronization in cat SI varies with stimulus properties (Roy and Alloway 1999), while local synchronization in primate SII depends on the animal s state of attention (Steinmetz et al. 2000). On the basis of these findings, we propose that the strength and spatial extent of long-range synchronization in SI and SII varies with the perceptual salience of tactile stimuli.

97 85 Chapter 4 Chapter 4. Coincidence detection or temporal integration? What the neurons in somatosensory cortex are doing 4.1. Introduction Neuronal synchronization is present in many brain regions during sensory stimulation, but its role in sensory processing is controversial (Gray 1999; Shadlen and Movshon 1999). Many investigators have proposed that neuronal synchronization is critical for transmitting sensory information, and have suggested that a major function of cortical neurons is to detect coincident events among their presynaptic inputs (Abeles 1982; Softky and Koch 1993; Alonso et al. 1996; Konig et al. 1996). While some evidence suggests that synchronous excitatory inputs are the predominant factor underlying the timing of cortical discharges (Softky and Koch 1993; Stevens and Zador 1998; Harsch and Robinson 2000; Salinas and Sejnowski 2000), other evidence suggests that cortical discharge behavior reflects how neurons integrate excitation and inhibition (Bernander et al. 1991; Reich et al. 1997; Troyer and Miller 1997; Shadlen and Newsome 1998). In fact, some have argued that mean firing rate must be more important for sensory coding than spike timing, because information about the timing of synaptic inputs is not preserved by integrate-and-fire mechanisms (Shadlen and Newsome 1994; Shadlen and Newsome 1998). Both sets of findings, however, are based almost entirely on studies that simulated presynaptic activity to study the responses of real or simulated neurons. Consequently, unless presynaptic inputs are directly monitored with respect to

98 86 the discharge behavior of cortical neurons in-vivo, much of the controversy surrounding neuronal synchronization and its relationship to cortical activity is unlikely to be resolved. We have previously shown that cutaneous stimulation causes synchronization among local groups of neurons in the ventrobasal thalamus (Alloway et al. 1995), but no study has examined how this synchronization affects the discharge behavior of target neurons in somatosensory cortex. Therefore, we extended our earlier work by testing the hypothesis that thalamic synchronization facilitates the firing of target neurons in the secondary somatosensory (SII) cortex. Furthermore, to determine the effective integration interval of SII cortical neurons, we measured their discharge probability with respect to the relative timing of synaptic inputs from different neurons in the ventrobasal thalamus. Recent results in the visual system indicate that near-coincident discharges among different neurons in the lateral geniculate nucleus increase the activation of neurons in striate cortex (Alonso et al. 1996; Usrey et al. 2000). Our results in the somatosensory system are consistent with those in the visual system, and compel us to suggest that neuronal synchronization is a fundamental mechanism for the transmission of sensory information from one brain region to another Materials and Methods Animal Preparation All procedures followed guidelines established by the National Institutes of Health on the use of laboratory animals, and these procedures were similar to those described previously (Johnson and Alloway 1996; Roy and Alloway 1999). Experiments

99 87 were conducted on two domestic cats in which a stainless steel recording chamber had been chronically implanted onto the cranium overlying the ectosylvian gyrus (SII cortex) and the ventrobasal thalamus during a sterile surgical operation. A bolt was also attached to the occipital ridge so that the head could be immobilized during neuronal recording. The animal was intubated through the oral cavity and was ventilated with a 2:1 gaseous mixture of nitrous oxide and oxygen containing 0.5% fluorothane to prevent reflexive movements. Because the animal's head was not held in a stereotaxic instrument, the concentration of isoflurane was lower than needed to anesthetize animals when the soft tissue surrounding the ears, eyes, and mouth was contacted by ear bars and other stereotaxic devices. Body temperature was maintained at 37?C, and both heart rate and end-tidal CO 2 were monitored continuously Electrophysiology Tungsten microelectrodes (2-4 M? ) were used to record extracellular discharges simultaneously from neurons in the ventrobasal complex and SII cortex. We recorded from SII cortex because it receives direct projections from the ventrobasal complex and has large receptive fields that are likely to overlap those recorded in the thalamus (Burton 1986). Two electrodes, with tips separated by microns, were advanced into the forelimb representation of the ventrobasal complex to record simultaneous discharges from pairs of thalamic neurons. With this electrode separation, discharges from one thalamic neuron were never recorded by both electrodes. Somatotopic maps of SII were used to guide placement of the cortical electrode into areas that matched the receptive field properties of the neurons recorded in the ventrobasal complex. Recording sites were

100 88 not confirmed by histology, but electrode depths were consistent with the location of neurons in layer IV or the deeper part of layer III. Extracellular neuronal waveforms were amplified, displayed, and converted into digital signals that were timestamped to a resolution of 0.1 ms (DataWave Technologies, Broomfield, CO). Timestamps were stored on hard disk for subsequent construction of PSTHs and CCGs Cutaneous stimulation Computer-controlled airjets were presented on 200 trials by means of hollow tubes oriented orthogonal to the hairy skin. Airflow was controlled by an electronic valve that was controlled by timestamped output from the data acquisition system (DataWave Technologies). Air pressure was maintained at 20 psi by a needle valve in series with a pressure gauge. Each airjet lasted 1 second or a total of 3 seconds for each trial. Airjets were separated by interstimulus intervals of 1 second, and each trial was separated by intertrial intervals of 4-6 seconds Cross-correlation analysis All CCGs and snowflake histograms were constructed from neuronal discharges recorded during airjet stimulation which lasted 600 seconds (3 seconds of stimulation times 200 trials). Spontaneous neuronal discharges occurring between airjet stimuli were not used in the construction of the CCGs or snowflake histograms. The CCGs and snowflake histograms were constructed according to previous descriptions (Perkel et al. 1967; Gerstein and Perkel 1972; Perkel et al. 1975; Johnson and Alloway 1996; Roy and Alloway 1999). The amount of correlated activity among pairs of simultaneously recorded thalamic neurons was determined by calculating the correlation coefficient for

101 89 events counted in the two adjacent bins surrounding time zero (Roy and Alloway 1999). Thalamocortical efficacy was calculated by counting the number of correlated events in the best 2 ms period of the raw CCG peak (appearing immediately after time zero) and then dividing this quantity by the number of reference events. To estimate stimulus-induced coordination we calculated a shift predictor (Gerstein and Perkel 1972; Perkel et al. 1975). The shift predictor represents the expected value of the CCG due to chance, and a peak in the raw CCG was considered statistically significant only if it exceeded the shift predictor by at least 1.96 standard deviations (i.e. beyond the 95% confidence limits). Because a standard deviation in the shift predictor is equal to its square root (Aertsen et al. 1989), we determined the 95% confidence limits by adding the shift predictor to the product of 1.96 times the square root of the shift predictor Results The impact of thalamic synchronization on cortical responsiveness was analyzed in 21 trios of neurons that were recorded from the cat somatosensory system. Each trio consisted of two neurons in the ventrobasal complex that were recorded simultaneously with a third neuron in SII cortex. The neurons had overlapping receptive fields on the hairy skin of the distal forelimb, and stimulus-induced responses were evoked by a pair of airjets administered, individually and in combination, to sites within these receptive fields. A typical example of one of these experiments is illustrated in Figure 4.1. To be included in our analysis, each thalamic neuron was required to have significant

102 90 Figure 4.1 Representative experiment (TC12) illustrating relationship between thalamic synchronization and thalamocortical coordination. A. Receptive fields for neurons in the ventrobasal complex (vb1, vb2) and secondary somatosensory cortex (SII). Red circles indicate airjet stimulation sites. B. Peristimulus histograms of neuronal responses to 200 trials of airjet stimulation. Binwidths, 25 ms. C. Crosscorrelograms (CCGs) displaying significant amounts of correlated activity as indicated by peaks exceeding the 95% confidence limits (red lines). Binwidths, 1 ms. D. Snowflake template showing how thalamocortical coordination in a neuronal trio is displayed. Dashed lines represent axes for displaying interneuronal interspike intervals (e.g. T SII minus T vb1 ). Red lines represent synchronous time axes for displaying instances in which two neurons discharge at the same time (e.g. T vb1 equals T vb2 ). As shown by the spikes trains and their corresponding points in the snowflake histogram, the time between synchronized thalamic events and subsequent cortical discharges is indicated by the distance from the central origin on the horizontal red axis. E. Snowflake histogram for experiment TC12. The white bins to the right of the origin indicates that the cortical neuron was most likely to discharge immediately after synchronous thalamic discharges (red arrow). Thalamocortical interactions for each thalamic neuron are indicated by faint diagonal bands (blue arrows). Color-coded legend and the number of neuronal discharges are shown to the left. Binwidths: 0.5 ms.

103 91 interactions with the cortical neuron as indicated by cross-correlation analysis (Perkel et al. 1967). As indicated by Figure 4.1C, correlated discharges in the thalamus and cortex were considered significant if they exceeded the 95% confidence limits at lag times (i.e. 0-4 ms) that were consistent with the conduction time of thalamocortical impulses (Yen et al. 1985; Johnson and Alloway 1996). In accord with our previous findings (Alloway et al. 1995), each pair of thalamic neurons displayed correlated activity during cutaneous stimulation. As shown in Figures 4.1 and 4.2, the amount of thalamic synchronization varied substantially among different neuronal trios, but was always characterized by a peak of correlated events at time zero of the CCG. Thalamic neurons with highly correlated activity had tall, narrow CCG peaks Figure 4.2 Cortical responses to varying amounts of thalamic synchronization. Top panel: snowflake histograms, illustrated as in Figure 4.1E, for three separate experiments (TC4, TC15, and TC16). The white bins on the horizontal time axis (to the right of the origin) indicate that cortical discharges were most likely after synchronous events in thalamic neurons VB1 and VB2. Binwidths: 0.5 ms. Bottom panel: raw CCGs indicating the proportion of thalamic activity that was synchronized in each experiment shown in the top panel. Correlation coefficients appear next to the CCG peak. Binwidths: 1.0 ms.

104 92 that were only 1 ms wide (e.g. TC15 and TC16). By comparison, thalamic neurons whose activity was weakly correlated had wider CCG peaks that transpired over 2-4 ms (e.g. TC4) Snowflake Analysis Snowflake histograms, which display the relative timing of discharges in 3 neurons (Gerstein and Perkel 1972; Perkel et al. 1975), revealed several findings about thalamic synchronization and its relationship to cortical responsiveness. High amounts of thalamic synchronization were evident by the presence of a prominent band of coordinated events on the horizontal time axis of the snowflake histogram (TC15 and TC16 in Figure 4.2). If the thalamic neurons were weakly correlated, however, the horizontal time axis was less prominent and had a fragmented appearance (TC12 and TC4 in Figures 4.1 and 4.2, respectively). Despite these variations in the amount of thalamic synchronization, the snowflake histograms indicated that cortical firing was highest immediately after instances in which both thalamic neurons discharged simultaneously. This pattern of coordination was represented by white bins on the horizontal time axis that were located 0-4 ms to the right of the central origin. Given that thalamic synchronization is represented by the horizontal time axis, the temporal displacement of the white bins from the central origin (ie. 0-4 ms) indicates the time lag between synchronous thalamic discharges and subsequent cortical discharges.

105 93 The time lag of cortical facilitation was consistent with the time lag of thalamocortical interactions. Thus, the part of the horizontal time axis representing the increase in cortical activity (i.e. the white bins) was usually intersected by diagonal bands depicting thalamocortical interactions (blue arrows in Figures 4.1 and 4.2). These diagonal bands represent instances in which thalamocortical interactions involving one thalamic neuron occurred independently of discharges in the other thalamic neuron, and these interactions usually involved time lags of 0-4 ms from the synchronous thalamocortical time axes (diagonal red lines in Figure 4.1D). As shown by the snowflake histograms, cortical responsiveness was most evident at the point where these diagonal bands intersected the horizontal time axis. Some snowflake histograms contained only one diagonal band even though the raw CCGs confirmed that both thalamic neurons had interactions with the cortical neuron. Diagonal bands were not apparent in these cases because thalamocortical interactions for one thalamic neuron rarely occurred independently of discharges in the other thalamic neuron. In experiment TC16, for example, the virtual lack of diagonal bands suggests that thalamocortical interactions for each thalamic neuron were tightly linked to synchronized activity with its thalamic partner (see Figure 4.2) Conditional Cross-correlation Analysis We developed a method, which we call conditional cross-correlation analysis, to measure the quantitative impact of thalamic synchronization on cortical responsiveness. As shown in Figure 4.3A, discharges in two thalamic spike trains can be sorted into three categories. Either both thalamic neurons discharge simultaneously or else one thalamic

106 94 Figure 4.3. Conditional cross-correlation analysis of synchronous and asynchronous thalamic discharges on neuronal responses in SII cortex. A. Procedure for classifying thalamic discharges as synchronous or asynchronous events. The center of each search interval served as the reference event for constructing a conditional CCG. B. Conditional CCGs of thalamocortical interactions in experiment TC15. Each CCG illustrates changes in the responsive of the SII neuron given that thalamic neurons VB1 and VB2 discharged synchronously or asynchronously at time zero. Thalamocortical efficacy is presented next to each CCG peak Binwidths: 1.0 ms.

107 95 neuron discharges by itself. After encountering a discharge in the spike train of either thalamic neuron, we examined the other spike train to determine if it also contained a discharge within a "search interval" of a specific duration. These events were sorted into different groups (i.e., VB1 and VB2, VB1 only, or VB2 only) that were correlated with the cortical spike train to generate separate CCGs for synchronous and asynchronous thalamic events. In cats, the ventrobasal thalamus and SII cortex have a serial relationship (Spreafico et al. 1981; Burton and Kopf 1984), and this allowed us to calculate an efficacy ratio which, in essence, indicates the probability that a thalamic event will evoke a subsequent cortical event (Levick et al. 1972; Aertsen et al. 1989). Representative results from using conditional cross-correlation analysis to analyze thalamocortical interactions in a neuronal trio are illustrated in Figure 4.3B. Regardless of the search interval duration, the probability of a cortical discharge was always higher when both thalamic neurons discharged within the search interval. Although thalamocortical efficacy declined when longer search intervals were used, prominent peaks were present in all of the CCGs based on synchronous thalamic events, but were not present in the asynchronous CCGs if the search intervals were 16 ms or longer. A comparison of the mean efficacy values obtained from all 21 neuronal trios revealed that synchronous thalamic events were substantially more effective for evoking cortical activity than asynchronous events. As Figure 4.4 illustrates, when the search interval was only 1.0 ms, the mean efficacy of synchronous thalamic events was twice as large as that produced by asynchronous events (12.4% vs. 5.9%). The efficacy of synchronous events gradually declined with longer search intervals and eventually

108 96 15 Synchronous Asynchronous Thalamocortical Efficacy (%) Search Interval (ms) Figure 4.4 Changes in thalamocortical efficacy as a function of search interval duration. Each line shows the mean (error bars indicate SEM) rate of synchronous and asynchronous thalamocortical efficacy that was calculated from the conditional CCGs of 21 neuronal trios. The horizontal cross-hatched bar represents thalamocortical efficacy (mean? SEM) obtained by applying conventional cross-correlation analysis to all 42 thalamocortical neuron pairs. became indistinguishable from the mean efficacy rate that was obtained by conventional cross-correlation analysis. The fact that progressively longer search intervals produced a systematic decline in thalamocortical efficacy indicates that cortical responsiveness is influenced by the relative timing of thalamic inputs Interspike Interval Analysis To determine more precisely the effective interval for cortical integration of different thalamic inputs, we modified our search strategy as shown by the diagram in Figure 4.5A. Thus, we searched for instances in which discharges in the two thalamic spike trains were separated by a specific range of intervals without any intervening discharges (i.e. an interneuronal interspike interval or ISI). In addition, an ISI search was considered successful only if both neurons failed to discharge during deadtime intervals

109 97 located immediately before and after the ISI. This last constraint reduced the possibility that postsynaptic effects produced by multiple discharges in the same thalamic neuron (homosynaptic integration) could summate with the postsynaptic effects of the discharges that define the ISI (heterosynaptic integration). After identifying instances in which specific ISIs were isolated from other discharges, these reference events were correlated with the cortical spike train to generate a conditional CCG. The results of this analysis are illustrated for a trio of neurons in Figure 4.5B. In this example, a constant search interval of 1 ms was used to identify instances in which the thalamic neurons discharged at a range of ISIs shown above each CCG. As these CCGs indicate, the cortical neuron had a 9.2% probability of discharging when both thalamic neurons discharged within a 1 ms period, but only discharged 4.4% of the time when the ISIs were 6-7 ms long. For longer ISIs, the probability of a cortical discharge remained relatively flat and did not decline any further (data not shown). The decline in thalamocortical efficacy to an asymptotic level indicates that this cortical neuron could integrate inputs from these thalamic neurons only if both neurons discharged within a 6 ms interval. To determine the effective interval for heterosynaptic integration, we identified neuronal trios that showed cooperative effects because these are the most likely to reflect direct, monosynaptic interactions. Neuronal cooperativity means that synchronous inputs cause the target neuron to discharge at a probability that is greater than would be predicted from the success rate of each presynaptic neuron alone. To understand this point, consider a pair of persons participating in the sport of skeet shooting. If each

110 98 Figure 4.5 Thalamocortical efficacy as function of interneuronal interspike intervals (ISIs). A. Diagram illustrating the procedure for identifying specific ISIs between discharges in the two thalamic neurons (VB1 and VB2). A search was considered successful only if no discharges occurred during the deadtime intervals. B. Conditional CCGs constructed from identifying specific ISIs in the thalamic spike trains of experiment TC12. A search interval of 1 ms was used to identify specific ranges of ISIs as shown above each CCG; deadtime intervals equal to the minimum ISI for each CCG. Percentages indicate thalamocortical efficacy. person alone has a 30% chance of hitting the target, then we could express this as a 70% failure rate (F). If both persons shoot simultaneously at the same target, the joint probability of a miss is (0.7) 2 or 49%. Hence, with both persons shooting simultaneously at the same target, the expected probability of a hit is 51% or 1 minus the joint probability of their individual failure rates (1 - F 2 ). In the case of thalamocortical interactions, the success rate of each thalamic neuron alone was given by the asynchronous efficacy that was obtained while using long search intervals (i.e. 128 ms as shown in Figure 4.3). This provided a basis for determining each thalamic neuron's independent failure rate, and

111 99 pairs of thalamic neurons were considered cooperative if their synchronous efficacy rate exceeded the predicted probability (i.e. 1-F 2 ). Among the 21 neuronal trios that were analyzed, we found 11 cases that displayed thalamocortical cooperativity. In these 11 cases, we initially searched for ISIs that were surrounded by equally long deadtime intervals. As indicated by Figure 4.6A, this searching paradigm indicated that mean thalamocortical efficacy was highest for synchronous thalamic discharges and gradually declined to an asymptotic level when the ISIs were 10 ms or longer. In response to synchronous thalamic events, the mean probability of a cortical discharge in these 11 neuronal trios was 12% or more than 3 times the efficacy level that was measured when ISIs were 10 ms or longer. Figure 4.6 Thalamocortical efficacy as a function of interneuronal ISIs. A. Mean thalamocortical efficacy (error bars indicate SEM) calculated from instances in which deadtime intervals were equal to the minimum ISI. B. Same as panel A except that deadtime intervals were 10 ms long when the ISIs were shorter than 10 ms. In both panels, search intervals were 1 ms long for ISIs ranging from 0-6 ms, but were gradually increased to 5 ms for the longest ISIs (15-30 ms) to increase the number of events available for conditional cross-correlation analysis.