Spatial summation in lateral geniculate nucleus and visual cortex

Transcription

1 Exp Brain Res (2000) 135: DOI /s RESEARCH NOTE Helen E. Jones Ian M. Andolina Nicola M. Oakely Penelope C. Murphy Adam M. Sillito Spatial summation in lateral geniculate nucleus and visual cortex Received: 21 July 2000 / Accepted: 28 August 2000 / Published online: 11 October 2000 Springer-Verlag 2000 Abstract We have compared the spatial summation characteristics of cells in the primary visual cortex with those of cells in the dorsal lateral geniculate nucleus (LGN) that provide the input to the cortex. We explored the influence of varying the diameter of a patch of grating centred over the receptive field and quantitatively determined the optimal summation diameter and the degree of surround suppression for cells at both levels of the visual system using the same stimulus parameters. The mean optimal summation size for LGN cells (0.90 ) was much smaller than that of cortical cells (3.58 ). Virtually all LGN cells exhibited strong surround suppression with a mean value of 74%±1.61% SEM for the population as a whole. This potent surround suppression in the cells providing the input to the cortex suggests that cortical cells must integrate their much larger summation fields from the low firing rates associated with the suppression plateau of the LGN cell responses. Our data suggest that the strongest input to cortical cells will arise from geniculate cells representing areas of visual space located at the borders of a visual stimulus. We suggest that analysis of response properties by patterns centred over the receptive fields of cells may give a misleading impression of the process of the representation. Analysis of pattern terminations or salient borders over the receptive field may provide much more insight into the processing algorithms involved in stimulus representation. Key words Spatial summation Surround suppression Receptive field Cat H.E. Jones ( ) I.M. Andolina N.M. Oakely P.C. Murphy A.M. Sillito Department of Visual Science, Institute of Ophthalmology, University College London, Bath Street, London EC1V 9EL, UK ams.admin@ucl.ac.uk Tel.: , Fax: P.C. Murphy Department of Physiology, St. George s Hospital Medical School, Cranmer Terrace, Tooting, London SW17 0RE, UK Introduction Varying patterns of spatial summation have been described in the visual cortex for a wide variety of visual stimuli (Blakemore and Tobin 1972; DeAngelis et al. 1994; Gilbert 1977; Henry et al. 1978; Hubel and Wiesel 1965; Kato et al. 1978; Knierim and Van Essen 1992; Li and Li 1994; Orban et al. 1979; Rose 1977; Sceniak et al. 1999; Schiller et al. 1976; Sengpiel et al. 1997; Sillito 1977; Sillito et al. 1995; Walker et al. 2000). The common feature is summation with increasing stimulus size within the classical receptive field (CRF) followed by a levelling in the response magnitude as size increases beyond the CRF, or in some cases a decline to a plateau level much lower than the optimal response. The decline is linked to the activation of suppressive mechanisms. These may be partially linked to the presence of discrete suppressive end-zones or side bands, or encompass a more diffuse mechanism overlying and surrounding the excitatory receptive field (e.g. DeAngelis et al. 1994; Hubel and Wiesel 1965; Kato et al. 1978; Orban et al. 1979; Walker et al. 1999). The interplay between the inhibitory and excitatory mechanisms characterising the fields of V1 cells has been explained in terms of a Difference of Gaussians model with unimodal, Gaussian inhibitory and excitatory fields broadly centred on the same location (e.g. Hawken and Parker 1987; Sceniak et al. 1999). Indeed a convenient way of quantifying the spatial summation characteristics of V1 cells is to compare their responses to varying the diameter of an optimally oriented circular patch of drifting grating, and a number of studies report data from this for both cat and primate (Sceniak et al. 1999; Sengpiel et al. 1997; Sillito et al. 1995; Walker et al. 2000). One can provide a similar description of the behaviour of the cells in the lateral geniculate nucleus providing the input to the visual cortex and this in turn would provide some insight into the manner in which the cortex integrates the geniculate input. Surprisingly, there is not a common body of data comparing the summation characteristics of LGN and visual cortical cells with the same stimuli and conditions.

2 280 Moreover, there are no published data quantifying the spatial summation characteristics of LGN cells with drifting grating stimuli of varying diameter. Here, we compare the spatial summation characteristics of LGN cells and visual cortical cells for the cat visual system. The data show the mean summation size of LGN cells to be much smaller than visual cortical cells, and, given the potent surround suppression characterising LGN cells, indicate that the cortical mechanism must integrate its much larger summation fields from the low firing rates associated with the plateau of the LGN cell responses. Our data also highlight the fact that the largest geniculate input will follow from the borders of the stimulus. Materials and methods Experiments were conducted on anaesthetised (70% N 2 O, 30% O 2, % halothane), paralysed (10 mg/kg/h gallamine triethiodide) female cats. End tidal CO 2, ECG waveform, intersystolic interval and EEG waveform were monitored at all times and the halothane level was adjusted to give an appropriate level of anaesthesia. Wound margins and ear canals were treated with local anaesthetic. The eyes were treated with atropine methonitrate and phenylephrine hydrochloride, protected with contact lenses, and brought to focus on a semi-opaque tangent screen/front surface mirror at a distance of either 0.57 or 0.46 m. All procedures were carried out in accordance with institutional and government regulations and the Principles of laboratory animal care (NIH publication No , revised 1985). Single unit activity was recorded from the A laminae of the LGN and from the primary visual cortex. Data were collected and visual stimuli generated using procedures described in detail elsewhere (Cudeiro and Sillito 1996). Recorded cells were hand mapped and classified before being subjected to a battery of computer-controlled visual tests to determine the receptive field type characterisation. Centring of the visual stimulus over the receptive field was assessed by exploring the spatial distribution of locations from which a contrast modulated patch, or a patch of optimally oriented (for cortical cells) drifting grating, elicited responses. A variety of patch sizes (0.2 2 ) were used. They were presented in a randomised sequence over a set of spatial locations defined in rectangular coordinates. The location giving the largest response was used to define the receptive field centre and the coordinates of our display adjusted accordingly. This involved several iterations, with variation of patch size and display coordinates to optimise centring. Spatial summation characteristics were assessed using stationary patches of drifting sinusoidal grating, with patch diameter being varied in a randomised sequence. LGN cells were studied with the same range of stimulus parameters used for cortical cells. Stimulus contrast was held at At the end of each penetration, key recording sites were marked with electrolytic lesions, so that the location of each cell could be accurately identified. Responses were computed from the mean firing rate after subtraction of the background discharge level. Spatial summation curves were plotted as a percentage of the optimal response along the ordinate against patch diameter (in degrees) along the abscissa. Surround suppression was quantified as the difference between optimal and plateau responses, expressed as a percentage of the optimal (0% no suppression, 100% complete abolition of response). Results Quantitative spatial summation curves were constructed for 112 LGN A laminae cells and for 90 cortical cells. The two populations were well matched in eccentricity (all were located within 12.5 of the area centralis; most were within ). Our LGN sample comprised 52 X cells, 55 Y cells and 5 inconclusively categorised cells. The cortical population included 44 simple cells, 42 complex cells and 4 unclassified cells. Spatial summation properties of LGN cells The spatial summation curve in Fig. 1A shows the response of an On-centre X cell. The cell s response increased with increasing patch diameter, reaching its maximal response for a 0.75 patch. Further increases in stimulus diameter led to a rapid decline in the cell s response. The response plateaued at a value of approximately 17% of the maximal response, giving a surround suppression index of 83% for this cell. A further example, from an Off-centre Y cell, is shown in Fig. 1B. It exhibited a very similar tuning profile. These tuning curves typified the profiles observed for most of the LGN cells tested. The average summation size was 0.90 ±0.03 SEM (n=112, range ). There was a small but statistically significant difference between the optimal summation diameters for X and Y cells (X cells, mean 0.81 ±0.04 SEM, n=52; Y cells, mean 0.98 ±0.04 SEM, n=55; P<0.005, Mann-Whitney U-test). The average degree of surround suppression observed across the group was 74%±1.61% SEM (range %). There was no significant difference between the surround suppression exhibited by X and Y cells. The variation in the distribution of spatial summation and surround suppression characteristics across our population of LGN cells is summarised by the filled bars in Fig. 1G, H. The histogram in Fig. 1G plots the distribution of optimal summation size and that in Fig. 1H plots the distribution of surround suppression. Spatial summation properties of cortical cells The spatial summation curves in Fig. 1C F document the range of response profiles observed in the visual cortex. Although many cortical cells exhibited surround suppressed tuning profiles, only a small minority of these exhibited tuning curves that mirrored those recorded in the LGN (Fig. 1C). Normally, even most of the strongly surround suppressed cortical cells exhibited broader spatial summation with larger optimal diameters than those seen in the LGN, as shown in Fig. 1D. At the other end of the cortical spectrum were cells that either reached a plateau level of firing without suppression (Fig. 1E) or continued to summate beyond the 12 maximum diameter of our display (not illustrated). The great majority of the cortical cells lay between these two extremes (Fig. 1F), with the entire range of possible summation sizes and degrees of suppression represented. The average summation size for the cortical population (n=90) was 4 times larger than that recorded for

3 281 Fig. 1A H Spatial summation characteristics of LGN and cortical cells. A F Representative spatial summation curves for two LGN cells (A, B) and four cortical cells (C F). Response magnitude (percentage of optimal ±1 SEM) is plotted against patch diameter (degrees). Cell type and degree of surround suppression is indicated on each graph. G, H Histograms showing the distribution of optimal summation size (G) and surround suppression (H) for LGN (filled bars) and cortical (open bars) cells. Cross hatching denotes cells where overlapping error bars for optimal and plateau responses were seen despite a surround suppression index >0, suggesting that the suppression was nonsignificant Fig. 2A, B Laminar distribution of cortical cell spatial summation characteristics. A Distribution of optimal summation size. B Distribution of surround suppression. Filled symbols denote cells where overlapping error bars for optimal and plateau responses were seen despite a surround suppression index >0

4 282 LGN cells at 3.58 ±0.27 SEM (range ). The average degree of surround suppression for the cortical cells was far lower than that seen in the LGN (mean 46%±3.36% SEM, n=90, range 0 100%). The variation in the distribution of spatial summation and surround suppression characteristics for the cortical cell population is summarised by the open bars in the histograms in Fig. 1G, H. The distribution of both parameters differed markedly from that of the LGN cell population. We explored the laminar distribution of optimal summation size (Fig. 2A) and surround suppression (Fig. 2B) for our population of cortical cells. There was a tendency for optimal summation size to increase with cortical depth. The average summation size was 2.23 (±0.18 SEM, n=37) for layer 2/3, 3.58 (±0.57 SEM, n=19) for layer 4, 4.50 (±0.48 SEM, n=21) for layer 5 and 5.96 (±1.01 SEM, n=13) for layer 6 cells. Although poorly and strongly surround suppressed cells were seen through all cortical layers, most of the strongly suppressed cells were found in the superficial layers. In contrast, layers 4 6 were dominated by cells with weaker suppression. Thus the average surround-suppression decreased with cortical depth, from 64% (±4.1% SEM, n=37) in layer 2 3, through 40% (±8.0% SEM, n=19) in layer 4, to 28% (±5.5% SEM, n=21) and 32% (±6.9% SEM, n=13) respectively in layers 5 and 6. Discussion The primary conclusion from these experiments is that the mean summation diameter (3.58 ) for cortical cells is substantially larger than for LGN cells (0.90 ). This might not be particularly interesting were it not for the fact that the LGN fields were so strongly suppressed by stimuli larger than their optimal diameter, with a mean suppression index of 74%. Although there was a significant difference in the optimal diameters of X and Y cells (0.81 and 0.98 respectively), the optimal for both groups was less than 1. Thus an optimal diameter stimulus for most cortical cells would strongly suppress the LGN cells providing input from the middle of the cortical cell s field. Because these observations were made at similar eccentricities and under the same stimulus and experimental conditions, there can be little doubt about the relative behaviour of the cortical and LGN cells. For our cortical cells the most strongly suppressed cells with the smallest fields were seen in layers 2/3 (Fig. 2). In particular, layers 4 and 6, which receive direct input from the LGN, were not characterised by small, strongly suppressed fields but rather included cells with summation diameters from 4 to 12. Given that the strength of the excitatory mechanism underlying the receptive fields of these cortical cells would fall into a Gaussian profile aligned over the centre of the receptive field, it is clear that the bulk of their geniculate input would be drawn from the suppressed plateau level of the LGN cells summation profiles. This has interesting implications for the issue of gain control, convergence and the relative weighting of the geniculate inputs. The strongest responses in the LGN would occur at the borders of the stimulus, where the receptive field centres of the LGN cells were engaged, but not all of the surround mechanism. In the central region of the cortical receptive field, with the strongest excitatory connections but the lowest LGN cell responses, it is interesting to speculate that the combination of large numbers of converging LGN cell inputs (up to 40 for a single spiny stellate cell in layer 4, Ahmed et al. 1994) may offset the low magnitude of the LGN cell responses to the optimal stimulus for the cortical cell. Driving the responses of cortical cells from the suppressed plateau of the LGN cell response may enable a higher level of gain control via the processes modulating the level of enhanced GABAergic surround suppression seen in the LGN (Francesconi et al. 1988; McCormick 1992; Sherman and Koch 1986; Sillito and Jones 1997; Sillito and Kemp 1983). It is also clear that corticofugal feedback to the LGN enhances surround suppression seen to moving stimuli (Murphy and Sillito 1987; Cudeiro and Sillito 1996). Additionally, for moving stimuli, the degree of synchronisation of LGN cell responses along the contour driving those cells with fields aligned along the axis of the cortical cell s optimal orientation may provide an effective amplification of the signal strength received by the cortical cell that is not simply reflected in the mean firing level of the individual LGN cells (see Sillito et al. 1994; Jones et al. 2000). It needs to be tested whether this synchronisation is enhanced or diminished when cells are responding in the suppressed plateau of their summation profile as opposed to their optimal focus. The available evidence suggests that layer 4 spiny stellate cells only summate inputs from separate LGN cells if the input spikes fall within a 7-ms window of each other (Usrey et al. 2000). An alternative hypothesis is that a substantial component of the cortical response may be drawn from geniculate inputs activated by the borders of the stimulus, where the LGN cell firing rate is significantly higher. One argument against this possibility is that, were it so, an annulus of drifting grating might be expected to be as effective a stimulus, or even more effective, than a central patch encompassing the centre of the receptive field. Clearly, this is not the case; thus one might presume that there has to be an integration of inputs from the plateau firing levels of LGN cells over the area of visual space integrated by the cell in question. However, the summation fields of cortical cells are generally far larger than those in the LGN, extending up to 12 in our data. Hence it seems clear that intracortical connections, such as those linking iso-orientation columns (e.g. Gilbert and Wiesel 1989; Kisvárday and Eysel 1992; Malach et al. 1993; Ts o et al. 1986; Ts o and Gilbert 1988), must contribute to this integration. From this perspective, it is worth noting that even layer 4 spiny stellate cells receiving direct geniculate input obtain less than 10% of their excitatory synaptic input from this route (Ahmed et al. 1994; Anderson et al. 1994; Peters and Payne 1993), the

5 remainder deriving from intracortical connections. Interestingly, intracortical connections have been shown to be sufficient to drive cortical cells that have been deprived of direct afferent input to their receptive field centres, through the production of either a real (Kaas et al. 1990; Gilbert and Wiesel 1992; Chino et al. 1992) or an artificial (Pettet and Gilbert 1992) scotoma. Moreover, there is evidence to suggest that geniculate afferents from beyond the CRF can also provide an effective drive when the cortical circuit is disrupted (Eysel and Schweigart 1999). These latter observations provide evidence for the existence of appropriate circuitry that could relay the enhanced inputs from the ends and borders of a stimulus. We suggest that cortical responses are underpinned by a combination of all three routes discussed above. Thus, finally, it appears that for the cat visual system we are left with a cortical mechanism that draws on geniculate inputs with small optimal summation sizes and strongly surround suppressed responses to create a range of different spatial summation patterns operating on a larger scale, with the addition of orientation selectivity. Surprisingly, strongly surround suppressed cortical cells with the smallest summation diameters reminiscent of LGN cell tuning profiles were most frequently observed in the superficial, output layers of the cortex. An additional point worth considering is that it may be the geniculate and cortical foci representing the borders of patterns that matter because this is where the excitatory drive is highlighted. Thus analysis by patterns centred on the receptive field of cells may give a misleading impression of the process of the representation. Analysis of pattern terminations or salient borders over the receptive field may provide much more insight. Acknowledgements This work was supported by the MRC. I.M.A. is supported by a Wellcome Trust 4-year PhD Neuroscience Scholarship. We are indebted to Ms. K. Lench for skilled technical assistance. References Ahmed B, Anderson JC, Douglas RJ, Martin KAC, Nelson JC (1994) Polyneuronal innervation of spiny stellate neurons in cat visual cortex. J Comp Neurol 341:39 49 Anderson JC, Douglas RJ, Martin KAC, Nelson JC (1994) Map of the synapses formed with the dendrites of spiny stellate neurons of cat visual cortex. J Comp Neurol 341:25 38 Blakemore C, Tobin EA (1972) Lateral inhibition between orientation detectors in the cat s visual cortex. Exp Brain Res 15: Chino YM, Kaas JH, Smith EL III, Langston AL, Cheng H (1992) Rapid reorganization of cortical maps in adult cats following restricted deafferentation in retina. Vision Res 32: Cudeiro J, Sillito AM (1996) Spatial frequency tuning of orientation-discontinuity-sensitive corticofugal feedback to the cat lateral geniculate nucleus. J Physiol (Lond) 490: DeAngelis GC, Freeman RD, Ohzawa I (1994) Length and width tuning of neurons in the cat s primary visual cortex. J Neurophysiol 71: Eysel UT, Schweigart G (1999) Increased receptive field size in the surround of chronic lesions in the adult cat visual cortex. Cereb Cortex 9: Francesconi W, Müller CM, Singer W (1988) Cholinergic mechanisms in the reticular control of transmission in the cat lateral geniculate nucleus. J Neurophysiol 59: Gilbert CD (1977) Laminar differences in receptive field properties of cells in cat primary visual cortex. J Physiol (Lond) 268: Gilbert CD, Wiesel TN (1989) Columnar specificity of intrinsic horizonal and corticocortical connections in cat visual cortex. J Neurosci 9: Gilbert CD, Wiesel TN (1992) Receptive field dynamics in adult primary visual cortex. Nature 356: Hawken MJ, Parker AJ (1987) Spatial properties of neurons in the monkey striate cortex. Proc R Soc Lond 231: Henry GH, Goodwin AW, Bishop PO (1978) Spatial summation of responses in receptive fields of single cells in cat striate cortex. Exp Brain Res 32: Hubel DH, Wiesel TN (1965) Receptive fields and functional architecture in two nonstriate visual areas (18 and 19) of the cat. J Neurophysiol 28: Jones HE, Andolina IM, Wei W, Gerstein GL, Sillito AM (2000) High orientation selectivity in the response synchronisation of LGN cells. Invest Ophthalmol Vis Sci 41:S51 Kaas JH, Krubitzer LA, Chino YM, Langston AL, Polley EH, Blair N (1990) Reorganization of retinotopic cortical maps in adult mammals after lesions of the retina. Science 248: Kato H, Bishop PO, Orban GA (1978) Hypercomplex and simple/complex cell classifications in cat striate cortex. J Neurophysiol 14: Kisvárday ZF, Eysel UT (1992) Cellular organization of reciprocal patchy networks in layer III of cat visual cortex (area 17). Neuroscience 46: Knierim JJ, Van Essen DC (1992) Neuronal responses to static texture patterns in area V1 of the alert macaque monkey. J Neurophysiol 67: Li C-Y, Li W (1994) Extensive integration field beyond the classical receptive field of cat s striate cortical neurons classification and tuning properties. Vision Res 34: Malach R, Amir Y, Harel M, Grinvald A (1993) Relationship between intrinsic connections and functional architecture revealed by optical imaging and in vivo targeted biocytin injections in primate striate cortex. Proc Natl Acad Sci USA 90: McCormick DA (1992) Neurotransmitter actions in the thalamus and cerebral cortex and their role in neuromodulation of thalamocortical activity. Prog Neurobiol 39: Murphy PC, Sillito AM (1987) Corticofugal feedback influences the generation of length tuning in the visual pathway. Nature 329: Orban GA, Kato H, Bishop PO (1979) End-zone region in receptive fields of hypercomplex and other striate neurons in the cat. J Neurophysiol 42: Peters A, Payne BR (1993) Numerical relationships between geniculocortical afferents and pyramidal cell modules in cat primary visual cortex. Cereb Cortex 3:69 78 Pettet MW, Gilbert CD (1992) Dynamic changes in receptive-field size in cat primary visual cortex. Proc Natl Acad Sci USA 89: Rose D (1977) Responses of single units in cat visual cortex to moving bars of light as a function of bar length. J Physiol (Lond) 271:1 23 Sceniak MP, Ringach DL, Hawken MJ, Shapley R (1999) Contrast s effect on spatial summation by macaque V1 neurons. Nat Neurosci 2: Schiller PH, Finlay BL, Volman S (1976) Quantitative studies of single-cell properties in monkey striate cortex. I. Spatiotemporal organization of receptive fields. J Neurophysiol 39: Sengpiel F, Sen A, Blakemore C (1997) Characteristics of surround inhibition in cat area 17. Exp Brain Res 116: Sherman SM, Koch C (1986) The control of retinogeniculate transmission in the mammalian lateral geniculate nucleus. Exp Brain Res 63:1 20

6 284 Sillito AM (1977) The spatial extent of excitatory and inhibitory zones in the receptive field of superficial layer hypercomplex cells. J Physiol (Lond) 273: Sillito AM, Jones HE (1997) Functional organization influencing neurotransmission in the lateral geniculate nucleus. In: Steriade M, Jones EG, McCormick DA (eds) Thalamus, vol 2: experimental and clinical aspects. Elsevier, Amsterdam, pp 1 52 Sillito AM, Kemp JA, Berardi N (1983) The cholinergic influence on the function of the cat dorsal lateral geniculate nucleus (dlgn). Brain Res 280: Sillito AM, Jones HE, Gerstein GL, West DC (1994) Featurelinked synchronization of thalamic relay cell firing induced by feedback from the visual cortex. Nature 369: Sillito AM, Grieve KL, Jones HE, Cudeiro J, Davis J (1995) Visual cortical mechanisms detecting focal orientation discontinuities. Nature 378: Ts o DY, Gilbert CD (1988) The organization of chromatic and spatial interactions in the primate striate cortex. J Neurosci 8: Ts o DY, Gilbert CD, Wiesel TN (1986) Relationships between horizontal interactions and functional architecture in cat striate cortex as revealed by cross-correlation analysis. J Neurosci 6: Usrey WM, Alonso JM, Reid RC (2000) Synaptic interactions between thalamic inputs to simple cells in cat visual cortex. J Neurosci 20: Walker GA, Ohzawa I, Freeman RD (1999) Asymmetric suppression outside the classical receptive field of the visual cortex. J Neurosci 19: Walker GA, Ohzawa I, Freeman RD (2000) Suppression outside the cortical classical receptive field. Vis Neurosci 17: