TEMPORALLY SPECIFIC BLOCKING: TEST OF A COMPUTATIONAL MODEL A Senior Honors Thesis Presented By Vanessa E. Castagna June 999 999 by Vanessa E. Castagna
ABSTRACT TEMPORALLY SPECIFIC BLOCKING: A TEST OF A COMPUTATIONAL MODEL Vanessa E. Castagna (John Moore, Thesis Chair), Psychology, University of Massachusetts, Amherst, MA This study investigated whether temporally specific blocking can be demonstrated in rabbits with a conditioned eyeblink response preparation. Temporally specific blocking asserts that blocking is confined to a particular CS-US interval. This study s prediction is derived from real time computational models of classical conditioning. Real-time computational models of conditioning represent time as a complete serial compound, a condition where a CS-US interval is divided into segments. Each segment is regarded as a separate and distinct CS. Real time models take into account the duration of stimuli and thus predict that blocking to a stimulus will be specific to a particular time interval. The results demonstrated blocking, but the blocking was not temporally specific. These results are contrary to the model s predictions. Real-time models may require extensive revision to accommodate these findings.
Temporally specific blocking: Test of a computational model Introduction The purpose of this study was to demonstrate temporally specific blocking of the conditioned eyeblink response in rabbits using Kamin s -stage protocol. In 99, Leon Kamin demonstrated blocking in rats in a conditioned suppression task. Kamin s experiment consisted of three stages. Two conditioned stimuli (CS, CS) were used with a grid shock as the unconditioned stimulus (US). In the first stage, CS was paired with the US, which produced conditioned suppression to CS. In stage, Kamin presented CS with CS. The two CSs were coextensive in time, and the US was presented as in stage. In the final stage of this conditioned suppression task, Kamin presented CS and CS alone without the US. Although conditioned suppression to CS was still intact, the rats failed to demonstrate conditioned responding when CS was presented alone. In effect, previous training to CS blocked CS from eliciting conditioned suppression. Marchant and Moore (9) demonstrated blocking of the conditioned nictitating membrane response in rabbits, a component of the conditioned eyeblink response. Experiment of Marchant and Moore (9) was an attempt to block a light CS with a tone CS. Five groups of rabbits were trained in stages. There was experimental group and control groups. In stage, the experimental group was presented with a tone paired with an eyeshock (US). Stage consisted of concurrent presentations of the tone
and light with the US. Stage consisted of light and tone presentations alone without pairings with the US. Rabbits in control group were placed in the training context, but they did not experience the CS or US. They then received the same training as the experimental group during stages and. Control group received concurrent tone and light presentations paired with the US during stage, tone and US pairings during stage, and the same extinction protocol in stage as the experimental group. Control group received tone and US pairings during stage, light and US pairings during stage, and the extinction protocol in stage. Like control group, control group did not experience a CS or US in stage. This group received light paired with the US in stage and the extinction protocol in stage. Refer to Figure. for a summary of these procedures. Results demonstrated a powerful blocking effect of the light by the tone, in that group gave only CR to the light during extinction whereas groups - gave between and CRs to the light. Using the same basic procedures as experiment, Marchant and Moore (9) next attempted to block the tone by the light. Eight rabbits were divided into groups of ; a control group and an experimental group. In stage, the experimental group was presented with a light paired with an eyeshock (US). Stage consisted of concurrent presentations of the tone and light with the US. Stage consisted of light and tone presentations alone without pairings with the US. Rabbits in the control group were placed in the training context, but did not experience the CS or US. They then received the same training as the experimental group during stages and. Results were not as successful as those of experiment, in that the experimental group gave a total of
Figure.: Protocol for experiment of Marchant and Moore, 9. Figure.: Protocol for experiment of Barnet et al., 99...
CRs to the tone during the extinction stage while very few CRs should have been evident if blocking occurred. The significant blocking effect seen in experiment was not demonstrated in experiment. Experiment of Marchant and Moore (9) was conducted on the basis of Rescorla and Wagner s (9) theory of conditioning. The theory suggested that conditioning to the light in stage failed to reach the asymptote by the end of the stage, thereby accounting for the failure of a significant blocking effect of the tone by the light. Therefore, in experiment of Marchant and Moore (9) the experimental group received twice as much training to the light during stage than did the experimental group in experiment. Eight rabbits were divided into groups of, group being the experimental group and group the control group. Group received conditioning sessions to the light during stage, conditioning sessions to the TL compound in stage, and extinction sessions where the T and L were each presented separately in the absence of the US. Group sat during stage training and received the same training as group during stage and extinction. Results showed that the experimental group yielded only CRs to the tone during the extinction session and group a total of 9. The difference in the number of CRs between the two groups was significant. Consequently, blocking of the tone by the light was demonstrated in agreement with the Rescorla-Wagner model. If blocking is temporally specific, such that CS-US pairings instill knowledge of when the US occurs after CS onset, then altering the CS-US interval from stage to stage ought to attenuate blocking. Several studies of blocking have examined the effect of
changing CS-US intervals (ISI) in the second stages of training from those employed in the first stages (Barnet, Grahame, & Miller, 99; Kohler & Ayres, 99; Maleske & Frey, 99). Kohler and Ayres (99) attempted to demonstrate that changing the timing relationships between the CS and US in the first stages of training would attenuate blocking in a head-poking task with rats. In stage, CS was of a fixed duration and terminated with a US. In stage, CS and CS were presented concurrently and paired with the US, but for some rats the CS duration and the ISI were fixed. For other rats, the CS duration and ISI varied from trial to trial. Stage was an extinction test used to assess conditioning to CS. The prediction was that CS presentations with varied durations would produce an element of surprise. Therefore, attenuation of blocking would be evident in rats presented with CSs with varied durations. Results showed a failure to demonstrate attenuation of blocking. Maleske and Frey (99) attempted to demonstrate that changing the ISI in stage from that in stage would attenuate blocking of eyeblink conditioning in rabbits. In stage, some groups received CS presentations of ms paired with a US at ms, and other groups received CS presentations of ms paired with a US at ms. During stage, all groups received training to a CS compound of ms paired with a US at ms. To assess blocking, stage consisted of ms CS presentations paired with a US at ms. A comparison was made of the CRs produced by groups trained with an ISI of ms during stage with the groups trained with an ISI of ms during stage. Results showed a failure in attenuation of the blocking effect.
Unlike Kohler and Ayres (99) and Maleske and Frey (99), Barnet et al. (99) did demonstrate attenuation of blocking by shifting the ISI between stages and of training. Barnet et al. (99) referred to this observation as evidence for a temporal encoding mechanism in associative learning, temporal encoding being the ability for an animal to learn the timing of the US by the onset of the CS, as in temporal specificity. In experiment, groups of rats underwent different schedules of training in a licksuppression task. During stage, each group was trained to, independently reinforced CSs. Group (F-Fe) received concurrent presentations of CS and the US, and forward pairings of CS and US. In this instance forward pairing indicated that US onset did not occur until after CS onset. Group (S-Fe) received simultaneous presentations of CS and US and forward pairings of CS and US. During stage, both groups received a compound CS of CS and a new CS paired in a forward relationship with the US. Refer to Figure. for a summary of this training schedule. Results demonstrated more blocking when pre-training occurred in a forward relationship between CS and US than in a simultaneous relationship. Thus, attenuation of blocking was evident when the ISI was shifted between stages and. Recently, Weidemann, Georgilas, and Kehoe (999) demonstrated evidence for temporal specificity in rabbit eyeblink conditioning using positive and negative patterning training protocols. In stage, training of the positive patterning task involved nonreinforced presentations of A alone (A-) and X alone (X-), and reinforced presentations of a serial compound CS (AX+). There were groups of rabbits that received training which differed in the interval between A and X on AX+ trials during stage ; either,
,, or. Stage was the same as stage, except all rabbits also received presentations of the non-reinforced serial compound CS (AX-) at each of the intervals,,,, and. Results showed that for each group, the highest level of responding to AX- during stage occurred where the A-X interval was the same as that used in stage of training. In stage of the negative patterning task, training involved reinforced presentations of A (A+) and X (X+), and non-reinforced presentations of the serial compound CS (AX-). There were groups of rabbits that received training which differed in the interval between A and X on AX- trials during stage ; either,,, or. Stage was the same as stage, except all rabbits also received presentations of the non-reinforced serial compound CS (AX-) at each of the intervals,,,, and. Results demonstrated that for each group, the lowest level of responding during stage occurred on trials where the A-X interval was the same as that used in stage of training. In sum, these experiments provided evidence that the temporal specificity of an ISI is a determinant in the production or attenuation of conditioned responding (Weidemann et al. 999). The present study s prediction of temporally specific blocking was derived from real- time models in which time is represented as a complete serial compound, a condition where the CS-US interval is divided into segments and each segment is regarded as a separate and distinct CS (Sutton & Barto, 99). Real-time computational models of conditioning take into account the duration of stimuli and thus predict that blocking to a stimulus will be specific to a particular time interval. A real-time model of 9
conditioning was first established by Moore and Stickney (9). This model was supplanted by the VET model by Desmond and Moore (9), a model that generates realistic CRs. Finally, the TD model was created after experimenting with the VET and SB models (Sutton & Barto 9). The VET model could not generate higher-order conditioning and the SB model could not explain shifts in CR timing that occur when the CS-US interval is changed during training. The TD model incorporates the positive attributes of the two models as well as compensating for their deficiencies (Moore & Choi, 99). Temporally specific blocking asserts that blocking is confined to the particular CS-US interval employed in stage of Kamin s protocol. Except for Weidemann et al. (999), none of the previously mentioned studies were able to demonstrate this sort of temporal specificity. The current study took advantage of the fact that the mixture of CS-US intervals produces bimodal responses, each mode corresponding to an ISI used in training. When applied to a compound CS, this mixed-isi training would be expected to result in bimodal responding to each CS. For blocking, pre-training to CS at particular ISI would build up a unimodal response specific to that ISI. When combined with an added CS in stage, preliminary training should produce blocking specific to the ISI employed in stage and not the other ISI. Experiment I Experiment I attempted to demonstrate temporally specific blocking of a tone CS by a light CS. Stage- training to the light should attenuate the amplitude of CRs to the
tone in stage, relative to control animals that did not receive training in stage. Stage- training employed a -ms ISI, and stage employed a mixture of ISIs, ms and ms. Therefore, stage- training was expected to result in CRs to the light with one amplitude peak, and this peak should occur in the neighborhood of ms after CS onset. By the end of stage- training, however, the light CS should show two amplitude peaks, one at ms and another at ms after CS onset. By contrast, one or both of the amplitude peaks of CRs to the tone CS should be reduced in magnitude compared to controls, thereby demonstrating blocking. To the extent that blocking is specific to the stage- ISI, the amplitude peak at ms on tone trials should be attenuated in experimental animals, whereas the amplitude peak at ms should remain robust, and control animals should demonstrate robust bimodal responses on tone trials during stage- training. Method Subject and Apparatus: The subjects were naive albino rabbits, E, F, G, and H, weighing approximately kg each at the start of the experiment. The animals were run two at a time in individually ventilated and soundproofed file drawers while restrained in Plexiglas boxes. A potentiometer was mounted on each rabbit s head and connected to a nylon suture loop in each rabbit s right eyelid by a small metal hook and thread. Movement of the right eyelid produced a signal that was displayed on the monitor of a laboratory computer. The computer controlled the presentation of CSs and the US. Conditioned responses were measured on non-reinforced test trials, as detailed below
under Data Analysis. Each rabbit was run for sessions per day, days a week for a total of sessions per week. Each session lasted minutes. The light CS (CS) was ms in duration and was delivered by two 9 vdc incandescent lamps. The bulbs were behind transparent plastic screens positioned mm in front of the animal s head. The tone CS (CS) was ms in duration. The tone was 9 db SPL and delivered by a speaker mm in front of the animal s head. On sessions - of stage- training, the tone was reduced to db in the attempt to reduce overshadowing of the light by the tone, as it became evident that conditioning to the more salient 9 db tone was competing with conditioning to the light. For all other experiments, tone intensity remained db. The US consisted of a mild electric shock of. ma and ms in duration ( dc pulses of. ms duration at Hz) delivered by a Grass stimulator and constant current unit. Current was delivered through leads attached to 9 mm stainless steel wound clips, one posterior to the right eye, and the other in a ventral position. Procedure and Design: Two days prior to the conditioning sessions, the rabbits were habituated by being placed in the Plexiglas restrainers and set outside the file drawers for a half hour. One day prior to conditioning sessions, rabbits were placed in the Plexiglas restrainers, shaved around the right eyes, sutured, and wound clips were fastened around each rabbit s right eye. Rabbits were then placed in the file drawers, at a time, for a half- hour each for habituation. Following days of habituation, conditioning sessions began. Stage consisted of sessions spaced over days. Only
experimental and control animal were run together. Each session in stage consisted of trials of a delay-conditioning paradigm. On trials, CS was paired with the US occurring at ms after CS onset as shown in Figure.. On trials presented randomly, the CS was presented alone as shown in Figure.. Experimental rabbits E and G received stage- training and rabbits F and H were sit controls that remained in the file drawers during stage- sessions. Rabbits F and H did not receive the CS or US presentations during stage. Stage consisted of sessions spaced over days. All rabbits received stage- training. Each session consisted of trials made up of trial types. Trial type (TL) consisted of CS and CS presented concurrently for ms with the US presented at ms after CS onsets as shown in Figure.. Trial type (TL) consisted of CS and CS presented concurrently for ms with the US presented ms after CS onsets as shown in Figure.. CS and CS were presented concurrently on trial type (TL-) in the absence of the US as shown in Figure.. CS was presented for ms in the absence of the US on trial type (T-) as shown in Figure., and CS was presented for ms in the absence of the US on trial type (L-) as shown in Figure.. Each session in stage consisted of presentations each of trial types and, and presentations each of TL-, T-, and L- trials. The trials were presented in random order. In both stages and, the intertrial interval (ITI) varied randomly from to seconds. Data Analysis: Conditioned responses were digitized and displayed on the monitor of the laboratory computer. The system recorded the position of the upper right eyelid every ms from CS onset, such that each CR waveform was represented as a vector of data
Figures.-.: Protocol for experiment I of this study. Figures. and. represent stage- protocol. Figures.-. represent stage- protocol. Each CS was ms in duration, and each US was ms in duration. Figures are not drawn to scale. Light, US at ms Light.. Tone and Light, US at ms Tone and Light, US at ms.. Tone and Light Tone.. Light.
points. Waveforms for each trial type, TL-, T-, and L-, were averaged in blocks of sessions, including those which appeared as no response. Because there were presentations of each trial type per session, the averaged waveforms for sessions were typically based on trials. In addition, the standard error of eyelid positions was computed for each of the data points of the averaged waveforms. The averaged waveforms were used to assess blocking. If stage- training to light blocks conditioning to tone in stage, then CR waveforms on T- trials for experimental animals should have lower amplitudes than for sit controls. If the blocking is temporally specific, then the amplitude peak corresponding to the -ms ISI should be smaller than the amplitude peak corresponding to the -ms ISI. Since stage involved conditioning to light with a -ms ISI, any such difference between the -ms and -ms peaks should be more pronounced among the experimental animals than the sit control animals. Another method of assessing blocking involved comparing the contributions of tone and light to the averaged waveforms of TL- trials. To do this, the averaged waveform of T- trials was contrasted with that of TL- trials at each of the data points. The difference between the two waveforms at each data point was squared, and the sum of squared deviations across all data points was noted. The same comparison was made between the averaged L- trials and the averaged TL- trials. The ratio of squared deviations from L- trials to the sum of squared deviations from T- and L- trials was designated by the Greek letter α. A high value of α indicated that the TL- waveform received a large contribution from the tone component of the compound CS and a proportionally smaller contribution from the light component. Blocking of the tone by
the light implies a smaller α among experimental animals than the sit controls. Thus, α measured the proportion of the TL- waveform that could be predicted by the T- waveform, and ( - α) measured the proportion of the TL- waveform that could be predicted by the L- waveform. The foregoing analysis of the respective contributions of the tone and light to the CR waveforms observed on tone plus light trials assumes that the tone and light waveforms combine linearly. This assumption is consistent with real-time computational models that predict temporally specific blocking, such as the TD model. We assessed the validity of this assumption by measuring the degree to which the CR waveforms of TLtrials could be described by linearly combining (superimposing) the waveforms from T- and L- trials. Recalling that T- and L- are vectors, a computerized search determined the best-fitting linear combination of the two waveforms, by the expression: β [ α(t-) + ( - α)(l-) ] The parameter β is a scaling factor that was allowed to range from. to. in increments of. in the computerized search routine. The predicted waveform was compared with the observed TL- waveform, and the squared deviations were computed for each of the data points in the two vectors. Dividing each squared deviation by the corresponding standard error squared yielded a χ (chi-square) for each data point. The sum of χ s over the data points provided an overall measure of goodness-of-fit between the observed TL- waveform and that predicted by the best-fitting linear combination of T- and L- waveforms. The χ analysis also indicated precisely where the combined waveforms deviated from the observed TL- waveforms.
Results and Discussion Experimental rabbits E and G were giving robust CRs to the light CS by the end of stage- training. Results from stage demonstrated evidence of blocking, but the blocking was not temporally specific. Results of blocking can be seen in the α and ( - α) values shown in Tables.., but the waveforms of Figures.. show that the blocking is not temporally specific. Experimental animals E and G demonstrated a greater contribution of the light to the TL- waveform than did control animals F and H, thereby demonstrating blocking. This evidence can be seen in Tables. -. which show α and ( - α) parameter values demonstrating the contribution of the tone and light to the TL- waveforms for each animal in stage. Although experimental animals demonstrated a greater contribution of the light to the TL- waveform, the contribution rapidly decreased from session to. The ( - α) columns in Tables. -. show that the light contribution to the TLwaveforms for experimental and control animals decreased from sessions to. This rapid decrease in light contribution was attributed to stimulus competition which caused overshadowing of the light by the tone. Because of this competition, the tone intensity was reduced by db. This decrease in tone intensity increased the contribution of the light for the remaining sessions so light contribution of experimental animals to the TL- waveform equaled or was greater than the contribution of the tone. Figures. -. show the average waveforms on TL-, T-, and L- trials during sessions - of stage- training for each animal. These results demonstrated blocking
Tables.-.: Tables demonstrating patterns in α values for experimental animal E and control animal F of experiment I. The protocol column refers to parameter file names. Blgt represents stage- training and Bstage represents stage- training.. E: Experimental. F: Control Protocol Sessions α ( α) β Blgt - - - - Blgt - - - - Blgt - - - - Blgt - - - - Bstage -..9. Bstage -... Bstage -.9.. Bstage -...9 Bstage -...9 Bstage -... Protocol Sessions α ( α) β Blgt - - - - Blgt - - - - Blgt - - - - Blgt - - - - Bstage -... Bstage -.9.9. Bstage -.9.. Bstage -... Bstage -.9.9.9 Bstage -.9..
Tables.-.: Tables demonstrating patterns in α values for experimental animal G and control animal H of experiment I. The protocol column refers to parameter file names. Blgt represents stage- training and Bstage represents stage- training.. G: Experimental. H: Control Protocol Sessions α ( α) β Blgt - - - - Blgt - - - - Blgt - - - - Blgt - - - - Bstage -..9. Bstage -... Bstage -... Bstage -... Bstage -.9..9 Bstage -..9. Protocol Sessions α ( α) β Blgt - - - - Blgt - - - - Blgt - - - - Blgt - - - - Bstage -...9 Bstage -.9.. Bstage -.9.. Bstage -.9..9 Bstage -.9.. Bstage -...
Figures.-.: Graphs showing waveforms to T-, L-, TL-, and the predicted TLwaveforms of experimental animal E and control animal F of experiment I. T- depicts trial type, L- depicts trial type, TL depicts trial type, and the top waveforms depict the predicted waveforms. The predicted waveforms are superimposed on the TLwaveforms in the boxes labeled TL. In the top boxes, shaded areas depict deviations of predicted waveforms from actual waveforms. For waveforms of T-, L-, and TL- trials, the shaded areas depict the confidence interval, + standard error of the mean.
. EXPERIMENTAL BSTAGE: E Sessions - Squared Deviation:..9(.T- +.L-) chi : 9.9 chi crit :.9 Peak =.99 mm @ ms; Peak =. mm @ ms EYELID POSITION (MM) E TL- + SEM n = E L- + SEM n = Peak =.9 mm @ ms; Peak =. mm @ ms Peak =. mm @ ms; Peak =. mm @ ms E T- + SEM n = Peak =. mm @ ms; Peak =.9 mm @ ms TIME FROM CS ONSET (MS)
. CONTROL BSTAGE: F Sessions - Squared Deviation: 9..(.T- +.L-) chi : 9. chi crit :.9 Peak =. mm @ ms; Peak =. mm @ ms EYELID POSITION (MM) F TL- + SEM n = F L- + SEM n = Peak =. mm @ ms; Peak =.9 mm @ ms Peak =.9 mm @ ms; Peak =. mm @ 9 ms F T- + SEM n = Peak =. mm @ ms; Peak =.9 mm @ ms TIME FROM CS ONSET (MS)
Figures.-.: Graphs showing waveforms to T-, L-, TL- trials, and the predicted TL- waveform of experimental animal G and control animal H of experiment I. T- depicts trial type, L- depicts trial type, TL depicts trial type, and the top waveforms depictt the predicted waveforms. The predicted waveforms are superimposed on the TL- waveforms in the boxes labeled TL. In the top boxes, shaded areas depict deviations of predicted waveforms from actual waveforms. For waveforms of T-, L-, and TL- trials, the shaded areas depict the confidence interval, + standard error of the mean.
. EXPERIMENTAL BSTAGE: G Sessions - Squared Deviation: 9..(.T- +.L-) chi :. chi crit :.9 Peak =. mm @ ms; Peak =.9 mm @ ms EYELID POSITION (MM) G TL- + SEM n = G L- + SEM n = Peak =.9 mm @ ms; Peak =. mm @ ms Peak =.9 mm @ ms; Peak =. mm @ ms G T- + SEM n = Peak =. mm @ ms; Peak =. mm @ ms TIME FROM CS ONSET (MS)
. CONTROL BSTAGE: H Sessions - Squared Deviation:..9(.9T- +.L-) chi :. chi crit :.9 Peak =. mm @ ms; Peak =. mm @ ms EYELID POSITION (MM) H TL- + SEM Peak =.9 mm @ ms; Peak =. mm @ ms n = H L- + SEM Peak Amplitude =. n = H T- + SEM n = Peak =.9 mm @ ms; Peak =. mm @ ms TIME FROM CS ONSET (MS)
because the waveforms on T- trials of experimental animals had lower amplitudes than those of control animals. Blocking, however, was not temporally specific because the peak amplitude on T- trials of experimental animals occurred at the -ms ISI. Experimental animals demonstrated T- waveforms with peaks of almost equal amlitudes like those of control animals. If temporally specific blocking were present, the amplitude peak to the tone at ms would have been reduced for experimental animals in comparison with control animals. Pre-training to the light would have blocked conditioning to the tone at ms. This did not occur. Experiment II Experiment II attempted to demonstrate temporally specific blocking of the conditioned eyeblink response by modifying the stage- procedure used in experiment I. Stage- training was the same as that of experiment I except that the ISI in experiment was ms. Since ms is a more favorable ISI for conditioning, temporally specific blocking might be easier to demonstrate at ms because it is a less favorable ISI, assuming that stage- training results in robust CRs. Method Subject and Apparatus: The subjects were naive albino rabbits, J, K, L, and M, each weighing approximately kg each at the start of the experiment. The subjects were run
in the same apparatus as in experiment I. Each rabbit was run for sessions per day, days a week for a total of per week. Each session lasted minutes. Procedure and Design: Before conditioning sessions, subjects were habituated by the same procedures as in experiment I. Conditioning sessions began with stage which was the same as stage for experiment I except that the US occurred ms after CS onset on trials where the US was presented as shown in Figure.. Figure. shows the stage- trial type where the CS was presented alone. Rabbits J and K were the experimental animals and received training during stage. Rabbits L and M were sit controls. After sessions of stage- training, all rabbits were trained during stage. Stage- training consisted of sessions and sessions were run per day. Stage- training was the same as that in experiment I except that CS was db for all sessions. Results and Discussion The method of analyses for experiment II was the same used in experiment I. Rabbits J and K were giving robust CRs to the light CS by the end of stage- training. Results from stage demonstrated evidence of blocking, but the blocking was not temporally specific. Results of blocking can be seen in the α and ( - α) values shown in Tables.., but the waveforms of Figures.. show that the blocking is not temporally specific, and in fact that there is no blocking. Experimental animals J and K demonstrated a greater contribution of the light to the TL- waveform than did control animals L and M, thereby demonstrating blocking. This evidence can be seen in tables. -. which show α and ( - α) parameter values
Figures.-.: Protocol for stage of experiment II. CS is ms in duration, US is ms in duration. Figures not drawn to scale.. Light, US at ms. Light
Tables.-.: Tables demonstrating patterns in α values for experimental animal J and control animal M of experiment II. The protocol column refers to parameter file names. Blgt represents stage- training and Bstage represents stage- training.. J: Experimental. M: Control Protocol Sessions α ( α) β Blgt - - - - Blgt - - - - Blgt - - - - Blgt - - - - Bstage -... Bstage -... Bstage -... Bstage -... Bstage -... Bstage -.9.9. Protocol Sessions α ( α) β Blgt - - - - Blgt - - - - Blgt - - - - Blgt - - - - Bstage -.9.. Bstage -.9.. Bstage -..9. Bstage -.9.. Bstage -... Bstage -.9..
Table.-.: Tables demonstrating patterns in α values for experimental animal K and control animal L of experiment II. The protocol column refers to parameter file names. Blgt represents stage- training and Bstage represents stage- training.. K: Experimental. L: Control Protocol Sessions α ( α) β Blgt - - - - Blgt - - - - Blgt - - - - Blgt - - - - Bstage -..9. Bstage -.9.. Bstage -... Bstage -... Bstage -... Bstage -.9.. Protocol Sessions α ( α) β Blgt - - - - Blgt - - - - Blgt - - - - Blgt - - - - Bstage -... Bstage -... Bstage -... Bstage -.9.. Bstage -.9.. Bstage -...
Figures.-.: Graphs showing waveforms to T-, L-, TL- trials, and the predicted TL- waveform of experimental animal J and control animal M of experiment II. T- depicts trial type, L- depicts trial type, TL depicts trial type, and the top waveforms depict the predicted waveforms. The predicted waveforms are superimposed on the TL- waveforms in the boxes labeled TL. In the top boxes, shaded areas depict deviations of predicted waveforms from actual waveforms. For waveforms of T-, L-, and TL- trials, the shaded areas depict the confidence interval, + standard error of the mean. 9
. EXPERIMENTAL BSTAGE: J Sessions - Squared Deviation:..(.T- +.L-) chi :. chi crit :.9 Peak =.9 mm @ ms; Peak =. mm @ ms EYELID POSITION (MM) J TL- + SEM n = J L- + SEM n = Peak =.9 mm @ ms; Peak =. mm @ ms Peak Amplitude =. mm @ 9 ms J T- + SEM n = Peak =. mm @ ms; Peak =. mm @ ms TIME FROM CS ONSET (MS)
. CONTROL BSTAGE: M Sessions - Squared Deviation:..(.9T- +.L-) chi :. chi crit :.9 Peak =.9 mm @ ms; Peak =. mm @ ms EYELID POSITION (MM) M TL- + SEM n = M L- + SEM n = Peak =. mm @ ms; Peak =. mm @ ms Peak Amplitude =.9 mm @ ms M T- + SEM n = Peak =.9 mm @ ms; Peak =. mm @ ms TIME FROM CS ONSET (MS)
Figures.-.: Graphs showing waveforms to T-, L-, TL- trials, and the predicted TL- waveform of experimental animal K and control animal L of experiment II. T- depicts trial type, L- depicts trial type, TL depicts trial type, and the top waveforms depict the predicted waveforms. The predicted waveforms are superimposed on the TL- waveforms in the boxes labeled TL. In the top boxes, shaded areas depict deviations of predicted waveforms from actual waveforms. For waveforms of T-, L-, and TL- trials, the shaded areas depict the confidence interval, + standard error of the mean.
. EXPERIMENTAL BSTAGE : K Sessions - Squared Deviation:..(.T- +.L-) chi :. chi crit :.9 Peak =.9 mm @ ms; Peak =. mm @ 9 ms EYELID POSITION (MM) K TL- + SEM n = K L- + SEM n = Peak = 9.9 mm @ ms; Peak = 9. mm @ ms Peak =. mm @ ms; Peak =. mm @ ms K T- + SEM n = Peak =.9 mm @ ms; Peak =.9 mm @ 9 ms TIME FROM CS ONSET (MS)
. CONTROL BSTAGE: L Sessions - Squared Deviation:..(.9T- +.L-) chi :. Peak =.9 mm @ ms; Peak =. mm @ ms chi crit :.9 EYELID POSITION (MM) L TL- + SEM n = L L- + SEM n = Peak =. mm @ ms; Peak =.9 mm @ ms Peak Amplitude =. mm @ 9 ms L T- + SEM n = Peak =. mm @ ms; Peak =.9 mm @ ms TIME FROM CS ONSET (MS)
demonstrating the contribution of the tone and light to the TL- waveforms for each animal in stage. Although experimental animals demonstrated a greater contribution of the light to the TL- waveform than control animals, the contribution rapidly decreased as stage- training progressed. The same was true for control animals. The ( - α) columns in Tables. -. show that the percentage of the light contribution to TL- trials for animals decreased from sessions to. In contrast with results shown in Tables.., the average waveforms on T- trials did not demonstrate evidence of blocking because waveforms on T- trials for experimental animals were greater in amplitude than those for control animals. As shown in Figures. and., control animals L and M produced bimodal responding on TL- and T- trials with waveforms having peaks of almost equal amplitudes. Neither control animal demonstrated robust conditioning on L- trials. Experimental animal J also demonstrated little conditioning to the light, and experimental rabbit K gave bimodal responses of equal amplitudes on TL-, T-, and L- trials. These waveforms can be seen in Figures. and.. If temporally specific blocking were present, the amplitude peak to the tone at ms would have been reduced for experimental animals in comparison with control animals because pre-training to the light would have blocked conditioning to the tone at ms. This did not occur
Experiment III Experiment III attempted to demonstrate temporally specific blocking of the conditioned eyeblink response to the tone by the light by decreasing the amount of training during stage using procedures employed in experiment I. Results from experiment I indicated that CRs to CS attained their maximum amplitude around session and began decreasing over the final sessions of training. It was therefore possible that failure to observe temporally specific blocking in experiment I was due to a postasymptotic decrement in conditioning to the light because of too much stage- training. In the attempt to eliminate this post-asymptotic decrement in responding, the training duration of both stages was reduced; sessions for stage and sessions for stage. Predictions for temporally specific blocking were the same as those for experiment I. Method Subjects and Apparatus: The subjects were naive albino rabbits, N, O, P, and Q, weighing approximately kg each at the start of the experiment. The animals were run in the same apparatus used in experiments I and II. Each rabbit was run for sessions per day, days a week for a total of sessions per week. Each session lasted minutes. Procedure and Design: Prior to conditioning sessions, the subjects were habituated using the same procedure as in the previous experiments. Stage- training was the same as that in experiment I except that rabbits received sessions instead of. Rabbits N and Q were the experimental rabbits and received training during stage while rabbits O and P
were sit controls. Stage was the same as experiment II except that rabbits received only sessions instead of. Results and Discussion The method of analyses for experiment III was the same as that used for the previous experiments. Experimental animals N and Q were giving robust CRs to the light by the end of stage. Results from stage demonstrated evidence of blocking, but the blocking was not temporally specific. Results of blocking can be seen in the α and ( - α) values shown in Tables.., but the waveforms in Figures.. show that the blocking was not temporally specific. Experimental animals N and Q demonstrated a greater contribution of light to the TL- waveform than that of control animals O and P, thereby demonstrating blocking. This evidence can be seen in Tables.. which show α and ( - α ) parameter values demonstrating the contribution of the tone and light to the TL- waveforms for each animal in stage. Although experimental animals demonstrated a greater contribution of the light to the TL- waveform, the contribution of the light to the TL- waveform decreased with stage- training for all animals as in experiments I and II. Figures... show the average waveforms to TL-, T-, and L- trials during sessions - of stage- training for each animal. Results from experimental animal Q demonstrated evidence of blocking because the amplitude of the T- waveform was smaller than that of both control animals. In contrast, though, experimental animal N did
not demonstrate evidence of blocking because the amplitude of the T- waveform was greater than that of both controls animals. Results failed to show any evidence that blocking was temporally specific, because the peak amplitude on T- trials for all animals occurred at ms. If temporally specific blocking were present, the amplitude peak to the tone at ms would have been reduced
Tables.-.: Tables demonstrating patterns in α values for experimental animal N and control animal O of experiment III. The protocol column refers to parameter file names. Blgt represents stage- training and Bstage represents stage- training.. N: Experimental. O: Control Protocol Sessions α ( α) β Blgt - - - - Blgt - - - - Blgt - - - - Bstage -..9. Bstage -... Bstage -... Bstage -... Protocol Sessions α ( α) β Blgt - - - - Blgt - - - - Blgt - - - - Bstage -... Bstage -.9.. Bstage -... Bstage -...
Tables.-.: Tables demonstrating patterns in α values for experimental animal Q and control animal P of experiment III. The protocol column refers to parameter file names. Blgt represents stage- training and Bstage represents stage- training.. Q: Experimental. P: Control Protocol Sessions α ( α) β Blgt - - - - Blgt - - - - Blgt - - - - Bstage -..9. Bstage -... Bstage -... Bstage -... Protocol Sessions α ( α) β Blgt - - - - Blgt - - - - Blgt - - - - Bstage -... Bstage -... Bstage -... Bstage -...
Figures.-.: Graphs showing waveforms to T-, L-, TL- trials, and the predicted TL- waveform of experimental animal N and control animal O of experiment III. T- depicts trial type, L- depicts trial type, TL depicts trial type, and the top waveforms depict the predicted waveforms. The predicted waveforms are superimposed on the TL- waveforms in the boxes labeled TL. In the top boxes, shaded areas depict deviations of predicted waveforms from actual waveforms. For waveforms of T-, L-, and TL- trials, the shaded areas depict the confidence interval, + standard error of the mean.
. EXPERIMENTAL BSTAGE: N Sessions - Squared Deviation:..(.T- +.L-) chi :, chi crit :.9 Peak =.9 mm @ ms; Peak =. mm @ ms EYELID POSITION (MM) N TL- + SEM n = 9 N L- + SEM n = Peak =.99 mm @ ms; Peak =. mm @ ms Peak =. mm @ ms; Peak =. mm @ ms N T- + SEM n = Peak =.9 mm @ 9 ms; Peak =. mm @ ms TIME FROM CS ONSET (MS)
. CONTROL BSTAGE: O Sessions - Squared Deviation:..(.T- +.L-) chi : 9. chi crit :.9 Peak =. mm @ ms; Peak =.9 mm @ ms EYELID POSITION (MM) O TL- + SEM n = 9 O L- + SEM n = Peak =.9 mm @ ms; Peak =. mm @ 9 ms Peak =. mm @ ms; Peak =. mm @ 9 ms O T- + SEM n = Peak =.9 mm @ ms; Peak =. mm @ ms TIME FROM CS ONSET (MS)
Figures.-.: Graphs showing waveforms to T-, L-, TL- trials, and the predicted TL- waveform of experimental animal Q and control animal P of experiment III. T- depicts trial type, L- depicts trial type, TL depicts trial type, and the top waveforms depict the predicted waveforms. The predicted waveforms are superimposed on the TL- waveforms in the boxes labeled TL. In the top boxes, shaded areas depict deviations of predicted waveforms from actual waveforms. For waveforms of T-, L-, and TL- trials, the shaded areas depict the confidence interval, + standard error of the mean.
. EXPERIMENTAL BSTAGE: Q Sessions - Squared Deviation:..(.T- +.L-) chi : 9. chi crit :.9 Peak =. mm @ ms; Peak =. mm @ ms EYELID POSITION (MM) Q TL- + SEM n = Q L- + SEM n = Peak =. mm @ ms; Peak =. mm @ ms Peak =. mm @ ms; Peak =. mm @ ms Q T- + SEM n = Peak =. mm @ ms; Peak =. mm @ ms TIME FROM CS ONSET (MS)
. CONTROL BSTAGE: P Sessions - Squared Deviation:..(.T- +.L-) chi :. chi crit :.9 Peak =. mm @ 9 ms; Peak =.9 mm @ ms EYELID POSITION (MM) P TL- + SEM n = P L- + SEM n = Peak =. mm @ ms; Peak =.9 mm @ 9 ms Peak =. mm @ ms; Peak =. mm @ ms P T- + SEM n = Peak =. mm @ ms; Peak =.9 mm @ ms TIME FROM CS ONSET (MS)
for experimental animals in comparison with control animals because pre-training to the light would have blocked conditioning to the tone at ms. This did not occur. Experiment IV Experiment IV attempted to demonstrate temporally specific blocking of the conditioned eyeblink response to tone by a light CS of ms duration instead of ms as in experiments I-III. This change in method was designed to enhance temporal segregation of learning the ISIs employed in stage. Temporally specific blocking might be easier to demonstrate if the CSs do not extend beyond the ms ISI employed in stage (Kehoe & Weidemann, 999). Method Subject and Apparatus: The subjects were naive albino rabbits, R, S, T, and U, weighing approximately kg each at the start of the experiment. The rabbits were run in the same apparatus used for the previous experiments. Each rabbit was run for sessions per day, days a week. In total, sessions were run per week. Each session lasted minutes. Procedure and Design: Prior to conditioning sessions, rabbits were habituated using the same procedures as in the previous experiments. Conditioning sessions began with stage, which consisted of sessions. There were trials per session. On of these trials CS was presented for ms and the US was presented at CS offset as indicated in Figure.. The remaining trials were presented randomly. CS was presented in the
absence of the US on these trials as indicated in Figure.. Rabbits R and T were the experimental animals and received training during stage. Rabbits S and U were sit controls. Stage- training consisted of sessions, but only sessions to were analyzed because data files for sessions were lost. All rabbits received training during stage. Stage consisted of trial types. On trial type (TL) CS and CS were presented concurrently for ms and the US was presented ms after CS onset as shown in Figure.. Trial type (TL) was the same as trial type except the US was presented at the offset of CS presentation as indicated in Figure.. CS and CS were presented concurrently for ms on trial type (TL-) in the absence of the US as indicated in Figure.. CS was presented for ms in the absence of the US on trial type (T-) as indicated in Figure.. CS was presented for ms in the absence of the US on trial type (L-) as indicated in Figure.. Results and Discussion The method of analyses for experiment IV was the same as that used for the previous experiments. Experimental animals R and T were giving robust CRs to the light CS by the end of stage- training. Results from stage demonstrated evidence of blocking, but the blocking was not temporally specific. Results of blocking can be seen in the α and ( - α) values shown in Tables.., but the waveforms of Figures.. show that the blocking is not temporally specific.
Figures.-.: Protocol for experiment IV of this study. Figures. and. represent stage- protocol. Figures.-. represent stage- protocol. Each CS was ms in duration, and each US was ms in duration. Figures are not drawn to scale. Light, US at ms Light.. Tone and Light, US at ms Tone and Light, US at mg.. Tone and Light Tone.. Light.
Tables.-.: Tables demonstrating patterns in α values for experimental animal R and control animal S of experiment IV. The protocol column refers to parameter file names. Blgt represents stage- training and Bstage represents stage- training.. R: Experimental. S: Control Protocol Sessions α ( α) β Blgt - - - - Blgt - - - - Blgt - - - - Bstage -.9..9 Bstage -9... Bstage -... Protocol Sessions α ( α) β Blgt - - - - Blgt - - - - Blgt - - - - Bstage -... Bstage -9... Bstage -...9
Tables.-.: Tables demonstrating patterns in α values for experimental animal T and control animal U of experiment IV. The protocol column refers to parameter file names. Blgt represents stage- training and Bstage represents stage- training.. T: Experimental. U: Control Protocol Sessions α ( α) β Blgt - - - - Blgt - - - - Blgt - - - - Bstage -... Bstage -9... Bstage -... Protocol Sessions α ( α) β Blgt - - - - Blgt - - - - Blgt - - - - Bstage -... Bstage -9... Bstage -...9
Figures.-.: Graphs showing waveforms to T-, L-, TL- trials, and the predicted TL- waveform of experimental animal R and control animal S of experiment IV. T- depicts trial type, L- depicts trial type, TL depicts trial type, and the top waveforms depict the predicted waveforms. The predicted waveforms are superimposed on the TL- waveforms in the boxes labeled TL. In the top boxes, shaded areas depict deviations of predicted waveforms from actual waveforms. For waveforms of T-, L-, and TL- trials, the shaded areas depict the confidence interval, + standard error of the mean.
. EXPERIMENTAL BSTAGE: R Sessions - Squared Deviation:..(.T- +.L-) chi :. chi crit :.9 Peak =. mm @ ms; Peak =.999 mm @ ms EYELID POSITION (MM) R TL- + SEM n = 9 R L- + SEM n = 9 Peak =. mm @ ms; Peak =.999 mm @ ms Peak =.99 mm @ 9 ms; Peak =. mm @ ms R T- + SEM n = 9 Peak =.99 mm @ ms; Peak =. mm @ ms TIME FROM CS ONSET (MS)
. CONTROL BSTAGE: S Sessions - Squared Deviation:..9(.T- +.L-) chi :. chi crit :.9 Peak =. mm @ ms; Peak =. mm @ ms EYELID POSITION (MM) S TL- + SEM n = 9 S L- + SEM n = 9 Peak =.99 mm @ ms; Peak =. mm @ ms Peak =.9 mm @ ms; Peak =.9 mm @ ms S T- + SEM n = 9 Peak =.99 mm @ ms; Peak =. mm @ ms TIME FROM CS ONSET (MS)
Figures.-.: Graphs showing waveforms to T-, L-, TL- trials, and the predicted TL- waveform of experimental animal T and control animal U of experiment IV. T- depicts trial type, L- depicts trial type, TL depicts trial type, and the top waveforms depict the predicted waveforms. The predicted waveforms are superimposed on the TL- waveforms in the boxes labeled TL. In the top boxes, shaded areas depict deviations of predicted waveforms from actual waveforms. For waveforms of T-, L-, and TL- trials, the shaded areas depict the confidence interval, + standard error of the mean. 9
. EXPERIMENTAL BSTAGE: T Sessions - Squared Deviation:..(.T- +.L-) chi : 9. chi crit :.9 Peak =. mm @ 9 ms; Peak =. mm @ 9 ms EYELID POSITION (MM) T TL- + SEM n = T L- + SEM n = Peak =. mm @ ms; Peak =. mm @ ms Peak =. mm @ ms; Peak =. mm @ ms T T- + SEM n = Peak =. mm @ ms; Peak =. mm @ ms TIME FROM CS ONSET (MS)
. CONTROL BSTAGE: U Sessions - Squared Deviation:..9(.T- +.L-) chi :. chi crit :.9 Peak =.99 mm @ ms; Peak =.9 mm @ ms EYELID POSITION (MM) U TL- + SEM n = U L- + SEM n = Peak =. mm @ ms; Peak =. mm @ ms Peak =.9 mm @ ms; Peak =. mm @ ms U T- + SEM n = Peak =. mm @ ms; Peak =. mm @ ms TIME FROM CS ONSET (MS)
Experimental animals R and T demonstrated a greater contribution of the light to the TL- waveform than did control animals S and U, thereby demonstrating blocking. This evidence can be seen in Tables.. which show α and ( - α) parameter values demonstrating the contribution of the tone and light to TL- waveforms for each animal in stage. Although experimental animals demonstrated a greater contribution of the light to the TL- waveform, the contribution of the light to the TL- waveform decreased with stage- training for all animals as in experiments I III. Figures.. show the average waveforms to TL-, T-, and L- trials during sessions - of stage- training for each animal. In agreement with results shown in Tables.., the average waveforms demonstrated blocking of the tone by the light ms for experimental animals because the response peaks of the T- waveforms for experimental animals were of lesser amplitudes than the T- waveforms of control animals. Although blocking was present, temporally specific blocking was not. If temporally specific blocking were present, the amplitude peak to the tone at ms would have been reduced for experimental animals in comparison with control animals because pretraining to the light would have blocked conditioning to the tone at ms. For both experimental animals, the amplitude peak at ms on T- trials was equal to that at ms, a result also demonstrated by control animals, therefore, temporally specific blocking was not demonstrated.
Experiment V Experiment V attempted to demonstrate temporally specific blocking of the conditioned eyeblink response to the tone by the light with the same ms CSs used in experiment IV with a -ms ISI in stage as in experiment II. Because ms is a more favorable ISI, blocking of the tone by the light might be more evident at the less favorable ms ISI, assuming that stage- training results in robust CRs. Method Subject and Apparatus: The subjects were naive albino rabbits, V, W, X, and Y weighing approximately kg each at the start of the experiment. The rabbits were run in the same apparatus used for the previous experiments. Each rabbit was run for sessions per day, days a week. In total, sessions were run per week. Each session lasted minutes. Procedure and Design: Prior to conditioning sessions rabbits were habituated using the same procedures as in the previous experiments. Conditioning sessions began with stage, which was identical to that of experiment IV, except that the US was presented at ms ( ms after CS offset). Rabbits V and Y were the experimental animals and received training during stage and rabbits W and X were sit controls. Stage- training consisted of sessions and was identical to that of experiment IV except that all sessions were analyzed.