Neuronal activity in the mediodorsal and intralaminar nuclei of the dorsal thalamus during classical heart rate conditioning

Brain Research, 532 (1990) 211-221 Elsevier

211

BRES 16010

Neuronal activity in the mediodorsal and intralaminar nuclei of the dorsal thalamus during classical heart rate conditioning D . A . Powell 1"2'3, Karen L. Watson I and Shirley L. B u c h a n a n 1'2 lNeuroscience Laboratory, VA Medical Center, Columbia, SC 29201, 2Department of Psychology, University of South Carolina, Columbia, SC 29208, and 3Department of Neuropsychiatry and Behavioral Science, University of South Carolina School of Medicine, Columbia, SC 29208 (U.S.A.) (Accepted 15 May 1990) Key words: Learning; Neuronal plasticity; Mediodorsal nucleus; Intralaminar nuclei of the thalamus; Rabbit; Heart rate

Multiple unit activity (MUA) was recorded from chronically implanted electrodes in either the mediodorsal (MD) or the intralaminar (IL) nuclei of the dorsal thalamus in separate groups of rabbits during (a) habituation of the cardiac orienting reflex, (b) Pavlovian heart rate (HR) conditioning, and (c) extinction of the HR conditioned response (CR). Other animals with similar recording electrodes received explicitly unpaired presentations of the conditioned stimulus (CS) and unconditioned stimulus (US). The cardiac orienting reflex and the HR CR consisted of bradycardia. However, tone-evoked tachycardia was obtained in animals that received CS/US unpaired presentations. MUA evoked by the CS consisted of a short latency (20-40 ms) increase under all conditions, which reached its maximum 200-300 ms after CS onset. This response habituated greatly during tone-alone pretraining, but was considerably greater in the paired than unpaired group during the later trials of conditioning in animals with MD, but not IL, placements. Instead, a longer latency increase (>500 ms) in MUA occurred in the paired but not in the unpaired animals in the IL group. The MUA increases in both instances, including the early, short latency increase in the MD group, and the longer latency increase in the IL group, were trial-related, and declined to pretraining levels during extinction, indicating that these neuronal changes had an associative basis. These findings suggest that neuronal activity in both MD and IL is related to the early events involved in Pavlovian conditioning, but that the relative roles of these two closely related thalamic nuclei in associative learning must be somewhat different. INTRODUCTION We have previously d e m o n s t r a t e d that the midline prefrontal cortex (PFC) is a m a j o r focus for the heart rate ( H R ) decelerations and blood pressure (BP) depressor responses observed during classical conditioning in rabbits. These cardiovascular adjustments occur in response to a previously neutral stimulus (conditioned stimulus, CS) that precedes an aversive unconditioned stimulus (US), and can be contrasted with instrumental conditioning in which the presentation of the US (or reinforcer) is d e p e n d e n t upon the occurrence (or nonoccurrence) of an o p e r a n t response. The present research focused solely on the bradycardia elicited by classical conditioning contingencies. Lesions of the midline P F C dramatically attenuate the magnitude of the conditioned bradycardia normally o b t a i n e d during classical conditioning 3'4. Electrical stimulation of the midline P F C elicits inhibitory cardiovascular changes which are similar to those observed during classical conditioning 7 and neuronal activity in the midline prefrontal area e v o k e d by classical conditioning contingencies is correlated with the magnitude of conditioned b r a d y c a r d i a 13.

It has been suggested that the prefrontal cortex in subprimate species be defined as that part of the prefrontal area that receives the projections of the m e d i o d o r s a l ( M D ) nucleus of the thalamus 26. This includes the midline P F C , as described above, as well as a m o r e lateral region of the prefrontal cortex along the dorsal b o r d e r of the rhinal fissure. H o w e v e r , lesions of the latter, so-called agranular insular cortex, have little effect on conditioned b r a d y c a r d i a , although stimulation of this more lateral prefrontal a r e a also produces inhibitory cardiovascular changes 25. F u r t h e r , M U A r e c o r d e d from the agranular insular cortex is not strongly associated with conditioned b r a d y c a r d i a , unlike that described above for the midline P F C 14. H o w e v e r , M D has also long been implicated in learning and m e m o r y processes 17" 20~21,22,33. Cells in the lateral p o r t i o n of M D project to the midline P F C 8'11'15, which has, as noted, been demonstrated to be intimately involved in the e l a b o r a t i o n of conditioned bradycardia. A n i m p o r t a n t question thus concerns whether neuronal activity r e c o r d e d from M D during classical conditioning is also r e l a t e d to conditioned bradycardia. In addition, the i n t r a l a m i n a r (IL) nuclei of the thalamus, including the c e n t r o m e d i a n , paracentral

Correspondence: D.A. Powell, Neuroscience Laboratory (151A), VA Medical Center, Columbia, SC 29201, U.S.A. 0006-8993/90/$03.50 ~) 1990 Elsevier Science Publishers B.V. (Biomedical Division)

212 a n d t h e c e n t r o l a t e r a l n u c l e i , h a v e a s t r o n g p r o j e c t i o n to t h e p r e f r o n t a l c o r t e x 8"12'~6"28. T h e s e nuclei also p r o j e c t t o the neostriatum, including the caudate nucleus, p u t a m e n , g l o b u s p a l l i d u s , a n d n u c l e u s a c c u m b e n s ~'12"16"28. H o w ever, whether MD

p r o j e c t s to t h e s e nuclei is at t h e

p r e s e n t t i m e c o n t r o v e r s i a l t'~5. D u e to t h e similar p r o j e c t i o n s o f t h e s e t w o nuclei to t h e P F C ,

unit discharges were generated during specific time epochs of interest during pretraining, classical conditioning, and extinction by the online computer. The window discriminator was set for each animal so that the pre-tone baseline activity recorded was approximately 10-15 Hz. MUA and HR was displayed on an oscilloscope and acoustically monitored to ensure that any subtle movement artifacts or baseline shifts could be documented and thus be excluded from the data analysis.

we recorded

m u l t i p l e u n i t activity d u r i n g classical c o n d i t i o n i n g f r o m b o t h t h e I L a n d t h e M D nuclei in t h e p r e s e n t e x p e r i ments.

MATERIALS AND METHODS

Animals Experimentally naive New Zealand albino rabbits of both sexes were employed as experimental subjects. The animals were approximately 6 months old at the beginning of the study and weighed 2-3 kg on arrival from a local USDA-licensed supplier. The animals were housed individually in an animal facility accredited by the American Association for Accreditation of Laboratory Animal Care and were maintained on ad libitum food and water. A 07.00-19.00 light/dark cycle was employed; all animals were studied during the daylight portion of this cycle. Principles for the care and use of experimental animals, as outlined by the US Public Health Service, were strictly followed in these studies.

Surgery The animals were anesthetized with ketamine hydrochloride (55 mg/kg body weight, i.m.), supplemented with acepromazine malate (2.2 mg/kg, i.m.) and xylazine (3.0 mg/kg, i.m.). Electrolytically etched epoxalite insulated Tungsten microelectrodes (1-4/~m tip, impedance 1-2 M£2 at 1000 Hz) were implanted in either the MD or the IL of the dorsal thalamus under stereotaxic guidance. The coordinates for implantation of these electrodes varied from animal to animal with respect to the following coordinates: 2-4 mm posterior to bregma, 1-3 mm lateral to the midline suture, and 8.5-9.5 mm ventral to the dura 31. Leads from the recording electrode were connected to an amphenol socket, which was then secured to the skull with dental acrylic and stainless steel machine screws. After an injection of bicillin, each animal was allowed from 1-2 weeks post-surgical recovery, during which time it was observed daily and additional antibiotics and/or analgesics administered as needed.

Recording procedures During behavioral training the animals were restrained in a standard rabbit restrainer within a ventilated and sound-attenuating enclosure (Industrial Acoustics Co.). The conditioned stimulus was a 4-s, 1216-Hz, 75-db, re-20/t N/cm2, square wave tone generated by solid state audio-oscillator circuits. The tone was delivered through speakers located in the experimental chamber approximately 30 cm above the animal's head. The unconditioned stimulus was a paraorbital electric shock train consisting of a 250-ms, biphasic, 3-mA, 60-Hz stimulus, administered through 2 stainless steel wound clips positioned approximately 1 cm above and below the right eye. All experimental events were controlled by a Heath LSI 11 microcomputer. Electrocardiogram (ECG) and stimulus event markers were recorded on a Grass Model 7D polygraph. The ECG electrodes were stainless steel safety pins previously inserted over the right shoulder and left flank. As noted in our earlier studies 19"23, insertion of these pins does not produce any obvious discomfort or infection. MUA was conventionally amplified with a WPI Model DAM 80 amplifier. This signal was band-passed (0.5-10 kHz) and notch (60 Hz) filtered and subsequently fed into an amplitude-window discriminator (WPI Model 121) for signal analysis. Discriminable

Behavioral training The animals were first adapted to handling and restraint for 2 daily 30-min sessions. Subsequently, HR and neuronal activity were assessed during 2 consecutive daily training sessions. During the first day, animals received 10 presentations of the 4-s tone without the paraorbital electric shock US, to assess neuronal activity during elicitation and habituation of the cardiac orienting reflex (OR). This phase of training will be referred to as the 'pretraining' phase. This first phase was followed by four 10-trial blocks of classical conditioning, in which the tone CS was paired with the eyeshock US, with a CS/US interval of 4 s for separate groups of animals with either MD or IL placements (paired condition). To control for nonassociative changes in MUA and HR, a random sequence of 40 tones and eyeshocks was presented that were specifically unpaired in a second group of animals with either MD or IL placements (unpaired condition). Day-2 training involved 2 additional 10-trial training blocks that were identical to the paired and unpaired training during the first day. These 20 training trials were followed by three 10-trial blocks of tone-alone presentations to assess neuronal and cardiac changes during experimental extinction. During pretraining, classical conditioning, and extinction trial-blocks, the tones were delivered at intervals of 75, 90 or 105 s on a random basis with a mean of 90 s. During training the interval separating the tones and eyeshocks in the unpaired groups was half that of the tone-tone interval for the paired groups, viz., 37.5, 45.0, or 52.5 s on a random basis with a mean of 45 s. Thus, the animals in the unpaired groups received the same number of tones and eyeshocks over the same period of time as the paired groups, but the tones and shocks were specifically never paired together in the unpaired condition. The 10th trial of each block of tone trials served as a tone-alone 'test' trial, during which the shock US was not presented. Heart rate was recorded on these trials during training and extinction, but was recorded during trials 1, 5, and 10 of the pretraining phase to 'track' the habituation of the OR.

Histology At the conclusion of the behavioral observations, the recording site was marked by passing an anodal current of 2.0-3.0 mA through the recording electrode for 20-25 s. Approximately 24 h later the animal was killed with sodium pentobarbital (75 mg/kg, i.v.) and perfused transcardially with physiological saline, followed by 10% formalin. After further immersion fixation at 4 °C in a solution containing 2% DSMO and 10% glycerol27, the brain was blocked in the transverse plane and frozen serial sections were taken at 40 ktm through the dorsal thalamus. These sections were counter-stained with .thionin and the electrode tract and electrolytic lesion subsequently plotted on plates adapted from the atlas of Urban and Richard 31 using a Zeiss drawing tube and microscope.

Data reduction and statistical analysis Heart period (interbeat interval duration) was measured in ms during a 4-s pre-tone baseline period and during the tone (CS) period on pretraining trials 1, 5, and 10, and on each of the test trials during training or extinction. CS-evoked changes in heart period on each of these trials was calculated by subtracting the mean 4-s baseline heart period from the heart period for each of the IBis during the tone. For purposes of data presentation heart period was converted to heart rate in all instances. As in our previous studies (i.e., ref. 13), MUA was evaluated by compiling peristimulus histograms for selected trial blocks in bins of 20 or 100 ms over 1-s

213 periods immediately preceding and following tone onset. Z-score equivalents for each of the tone bins was calculated by subtracting the mean baseline discharge from the discharge evoked by the tone, and dividing this difference by the standard deviation of the baseline discharge. Both the heart period and neuronal data of the MD and IL groups were submitted to repeated measures analysis of variance (ANOVAs), generally involving factors of (a) condition (viz. paired vs. unpaired), (b) interval during tone presentation (viz. ms) and (c) trial (for HR) or trial-block (for MUA). Significant effects were further analyzed by post-hoc application of Duncan's Multiple Range Test.

RESULTS

Histology Fig. 1 shows electrode placements in the M D (closed circles) and IL (closed squares) nuclei on line drawings through 4 A P levels of the thalamus. Fig. 1A shows placements of animals that received paired CS/US training, while Fig. 1B shows similar placements in animals that received unpaired, randomly presented CS/US presentations. As can be seen there were no systematic differences between placements in the paired and unpaired conditions of either the MD or IL groups.

sisted of bradycardia that was greatest during the initial block of 10 trials, and declined somewhat over subsequent blocks of trials, but which was still considerably greater on these trial-blocks than during pretraining or extinction. The H R changes in the animals presented unpaired tones and paraorbital shocks, however, consisted of tachycardia, also as previously described 13'14. Fig. 3 illustrates these findings. This figure shows CS-evoked H R changes from pre-CS baseline as a function of postoCS IBis during test trials associated with the first 4 blocks of 10 trials of acquisition training. Also shown are the H R changes from pre-tone baseline for the last trial of pretraining and the last trial of extinction. These data are shown separately for the M D and IL placements and for the paired vs. unpaired conditions. A N O V A of these data revealed significant condition (viz., paired vs. unpaired) differences for both M D and

Heart rate: pretraining The initial cardiac response to the first tone during trial 1 of pretraining consisted of a dramatic decrease in H R , which reached its maximum magnitude by the 10th IBI, just prior to tone offset in most animals. However, this response had declined greatly in magnitude by trial 5, and had virtually disappeared by trial 10, representing virtually complete habituation of the cardiac OR. Figure 2 illustrates tone-evoked changes in H R in the paired and unpaired groups during the first, middle, and last trial of pretraining. A N O V A revealed no differences between the M D and IL groups. Thus, the H R data shown are pooled for the electrode placement dimension. These data are plotted in beats/min (BPM) change from pre-CS baseline as a function of 10 post-CS IBis. Separate A N O V A s of the H R data for the IL and MD groups revealed significant IBI and trial effects for both groups ( I B I : F9,17 l = 5.31; P < 0.0001, and F9,1o8 = 2.92, P < 0.003; and trial: F2.38 = 27.9, P < 0.0001 and F2,24 = 13.6, P < 0.0001 for, the M D and IL placement groups, respectively). However, the condition (viz. paired vs. unpaired) main effect was not significant nor were any condition interactions significant for either group. Thus, the H R response to the tone was comparable in the paired and unpaired conditions prior to the initiation of classical conditioning or unpaired CS/US training.

Heart rate., classical conditioning As has been found in previous studies 3'19"23, CSevoked H R changes during classical conditioning con-

'

Fig. 1. Electrode placements located in the mediodorsal nucleus (O) and in the intralaminar nuclei ( . ) of the thalamus. Drawings represent A-P levels that are approximately 2.0, 3.0, 4.0 and 5.0 mm posterior to bregma, based on plates adapted from the atlas of Urban and Richard 31. A shows placements in animals that received paired CS/US training and B shows similar placements in animals that received unpaired randomly presented CS/US presentations.

214 IL placements (MD: Fl,19 = 5.37, P < 0.03 and IL: Fl.12 = 8.3, P < 0.01), as well as a condition × IBI interaction (MD: F9,17~ = 8.6, P < 0.0001; IL: F9,ao8 = 7.8, P < 0.0001). Post-tests of the H R change in the paired conditions revealed that the greatest magnitude change, which occurred during the 10th IBI, at the end of the 4-s CS, was significantly larger on acquisition trial-block 1, compared to trial-blocks 2, 3, and 4. The latter were not significantly different from each other. All 4 acquisition trial-blocks were significantly different from both extinction and pretraining (largest P < 0.01). In the unpaired conditions, the tachycardia associated with the 10th IBI was significantly greater during the 4 training trial-blocks, compared to extinction and pretraining (largest P < 0.05), but the training trial-blocks did not differ significantly from each other.

Heart rate: day 2 training CS-evoked H R changes during training on day 2 were very similar to those during day 1. This response again consisted of conditioned H R decelerations, beginning at tone onset in both the IL and MD groups, which became greater over the tone interval to reach a maximum by the 10th interbeat interval. Also like session 1, both unpaired groups revealed H R increases from baseline during tone presentation. The paired vs. unpaired condition and IBI main effects, and the condition x IBI interaction, were significant for both the IL and MD analyses (largest P < 0.01). Heart rate: extinction Tone-evoked H R changes declined during extinction from those obtained during conditioning (see Fig. 3). However, these H R changes did not completely extin-

~

y -tO

TRIAL

•

0-~0 I1.

_ ~

_

- ~

-

~

- ~

_

~

- ~

-

,~

~

Multiple unit activity: pretraining Tone-evoked increases in M U A occurred in both the IL and MD placement groups during pre-training, regardless of whether the animals later received paired or unpaired CS/US training. However, the magnitude of the M U A increase, as well as its change over bins, was different for the two placement groups. Fig. 4 illustrates these changes in M U A during pretraining, separately for the paired and unpaired conditions with MD and IL placements. These data are Z-scores, as described above, shown as a function of ten 100-ms post-tone bins. Average M U A is shown separately for the first 3, second 3, and last 4 trials of pretraining. Tone-evoked M U A increases in MD showed a rapid augmentation over the first 3-4 100-ms bins, reaching a maximum Z-score of 6-7, after which M U A declined. This was true of all 3 stages of pretraining, but was most pronounced during trial-block 1, early during pretraining. In fact, the middle and last trial-blocks were considerably smaller than the first in both the paired and unpaired animals. ANOVA of these data for the MD groups revealed significant trial-block (F2,57 = 7.68, P < 0.001) and bin (F9,279 = 6.46, P < 0.0001) effects. However, the condition (viz.,

TRIAL10

-

m

guish. A N O V A revealed a significant change from baseline for all animals during extinction (largest P < 0.01). An A N O V A comparing the paired and unpaired conditions also revealed significant differences between the conditions for both the IL and MD placements (largest P < 0.01), again reflecting the different directions of H R change in the two conditions. However, as described above, the H R changes during extinction were significantly smaller than those during acquisition for both paired groups.

5

TRIAL1

A

~' ~' I ,' ,' ; '~ ~' ,'o

B i ~ ~ I ~ ~ ~ ~ ~,~

POST-TONEINTERBEATINTERVALS

Fig. 2. Heart rate change from pre-tone baseline in the (A) paired and (B) unpaired groups during the first, middle and last trial of pretraining in which the tone was presented without the paraorbital shock US.

215 UNPAIRED INTRALAMINAR NUCLEI

MEDIODORSAL NUCLEI B

~

*2O-

BJ = m

÷l(~-

~J Z 0 I-

~

~

~ "

~ ~ , ~

l~ ~'-,,P~_

"-~"~

'4 ' I

EXTINCTION

_ ~ - - ~ - - o EXTINCT,ON ~~ACOUlS,T,ON

-

~3 A C Q U I S I T I O N

RETRAININO

~

~-

R~---.,~,,.,,"='--~--I~-----~

1 PRETRAINING

PAIRED

~ ~ ~ ~ Z~ << Z~

:::::::

O-

O~

~

~

PRETRAINING ~

EXTINCTION

~~,~

-~o-

m

< ¢

~

-20 -

¢

<

~ ACQ~ ACQ

-30'

• PRETRAINING r~ E X T I N C T I O N

~ ~' ~ I g g ; ~ ~ ,'0 POST-TONE

I '~ ~ I INTERBEAT

~ ~ ; ;

,'o

INTERVALS

Fig. 3. Heart rate changes elicited during the last trial of each of 4 blocks of training trials compared with (A) the last trial of pretraining after the heart rate changes associated with orienting had habituated and (B) the last trial of extinction. A illustrates the heart rate changes elicited by the CS in the paired CS/US conditioning groups and B illustrates similar heart rate changes obtained in groups that received unpaired CS/US presentations. These data are shown as a function of interbeat intervals occurring after CS onset and are plotted as difference scores occurring during the first 10 interbeat intervals after CS onset. paired vs. unpaired) main effect was not significant (F1,31 = 1.3, P > 0.10). M U A increases in the IL placement group were more gradual and of smaller magnitude. The change over bins and decline over trials were nevertheless also significant for the IL placements (bin: F9,153 = 3.74, P < 0.0003, and trial-block: F2,3] = 8.15, P < 0.001). However, as with the M D analysis, the condition main effect was not 0.04, P < 0.10) for the IL placements. Post-test analysis revealed that changes during trial-block 1 were reliably greater than during trial-blocks 2 and 3 for both the M D and IL placements (P < 0.05), but the latter trial-blocks were not significantly different from each other. These analyses thus suggest that, like the H R response to the tone, tone-evoked increases in M U A in both M D and IL habituated as a function of pretraining exposure without the US. However, unlike the H R O R , it did not decline to pre-tone baseline levels in either significant

(Fl,17

=

5 illustrates CS-evoked M U A activity for successive 10-trial blocks during conditioning, compared with the last 10-trial block of extinction and the last 4-trial block

s.

MD UNPAIRED

MD PAIRED

• TRIALS 1 - 3 • TRIALS 4 - 6 • TRIALS 7 - 1 0

7. 6-

~ IM5~,. O = 3_ N ~ 2. t'-l. ~. it ~o ~~ ~ ,~ ~3 ~

•

,

,

'

'

,

'

'

,

,

INTRALAMINAR PAIRED

'

'

,

,

'

'

'

'

'

7

8

9

INTRALAMINAR UNPAIRED

group. 1

Multiple unit activity in the mediodorsal nucleus during classical conditioning There was a training-induced increase in evoked M U A within the initial block of training trials in both M D groups. As with pretraining, this response consisted of an initial increase in discharge, which peaked in most cases during the 3rd 100-ms period following tone onset. Fig.

2

3

4

5

6

7 100

8

9

10

1

MILLISECOND

2

3

4

5

6

10

INTERVALS

Fig. 4. Mean multiple unit discharge associated with MD (A) and IL (B) placements during pretraining. The data are plotted as a function of 100-ms intervals after CS onset in Z-score units normalized with respect to pre-CS (viz. baseline) discharge. The data are shown separately for the groups that later received paired CS/US (left) and unpaired CS/US (right) classical conditioning training. Data are also shown separately for different stages of pretraining.

216 of pretraining, for the paired and unpaired animals with MD placements. As with the pretraining MUA, described above, these data are shown as a function of ten 100-ms post-tone bins. As can be seen, increases in M U A were observed in both the paired and unpaired conditions. Peristimulus time histograms, in which 20-ms bins were employed, revealed that the latency to reach a significant Z-score varied between 20-40 ms in both the paired and unpaired conditions. Thus, animals in both conditions showed similar short-latency increases in M U A associated with CS onset. However, the pattern of discharge across post-CS 100-ms bins was different in the paired and unpaired groups. Whereas M U A remained elevated in the paired group across all 10 bins, it tended to decrease during bins 4-10 in the unpaired group, especially during later trial-blocks. Interesting differences also occurred across trial-blocks. Although there was a trial-block-related increase from trial-blocks 1-3 in the paired group, a trial-block-related decrease in M U A was observed in the unpaired group over all 4 trial-blocks. Thus, the smallest increase in M U A from pretraining levels occurred during trial-block 1 for the paired group, and this increase in M U A was dramatically augmented during the 2nd and 3rd trial-blocks. However, as was previously found in animals with prefrontal placements 13 there was a decline from trial-block 3 to trial-block 4,

such that CS-evoked M U A appeared to be similar during trial-blocks 1 and 4. Nevertheless, both were considerably elevated over that obtained during the pretraining and extinction conditions. On the other hand, the largest increase in M U A in the unpaired group occurred during the initial trial-block and this change declined systematically over subsequent trial-blocks. Consequently, the smallest increase in CS-evoked discharge occurred during trial-block 4 in the unpaired condition. ANOVA of these data revealed significant trial-block (F5,143 -- 4.8, P < 0.0004) and bin (F9.270 = 26.0, P < 0.0001) main effects. The paired vs. unpaired condition × bin (F9,270 = 2.36, P < 0.01) and paired vs. unpaired condition x bin × trial-block 1.9, P < 0.02) interactions were also significant. Post-test comparisons revealed that all bins during all 4 trial-blocks in the paired condition were significantly greater than those in either extinction or pretraining (largest P < 0.01). Moreover, bins 2-7 and bin 10 were significantly greater during trial-blocks 2 and 3 than during trial-blocks 1 and 4. However, the latter 2 trial-blocks were not significantly different from each other for any bin. In the unpaired condition, all bins of trial-blocks 1, 2, and 3 were significantly different from those during extinction and pretraining (largest P < 0.01), but during trial-block 4 they were significantly different only from those of the

(F45,1287=

)AIRED &

UNPAIRED B

8ACQ- 2 ACQ- 3 7~

ACQ- 1

~

6-

~N ~ ~

5--

~(~ ¢{ f-

4--

I~1

3--

ACQ- 4 ACQ- 1

ACQ- 2 ACQ-3 ACQ- 4

/

,~)~1A7 ~ ,

~

EXTINCTION

~ ~ ~_

,~ / ~ , ~ , ~ ',~,

~-~ ~

.

3

.

4

.

5

.

~

.

7

8

~

9

,

10 100 MILLISECOND

~

~ ~ 1 INTERVALS

~

~

~

~

~

1~

Fi~. 5. Mean CS-cvokcd multiple unit discharge associated with mcdiodorsaI thalamic placements, plotted as a function of ]00-ms intc~als after CS onset in Z-score units that wcrc normalized with respect to pr¢-CS (viz. baseline) discharge. A shows chanscs associated with the paired CS/US ~roup and B shows similar chan~es associated with the unpaired CS/US ~roup. Mean chan~c in M U A is shown (a) for each of the 4 Mocks of ]0 trainin~ trials on day ] (ACQ ]-4), (b) for the last 4 trials of prctraJnin~ and (c) for the last ]0 trial-blocks of extinction.

217

B-

~.1

o N

PAIRED

UNPAIRED

A

B

6-

~/0~

ACQ

4

ACQ- 3

51Z)~

ACQ- 2 ACQ- 1 EXTINCTION

4-

PRETRAINING

~

3o

PRETRAINING ACQ- 4 Aco-2

2"

'

/

~ACQ-3

x.

/~ O

~

~

ACQ- 1

1-

1.

. 2 . . 3 . 4. . 5. . 6

7

8

9

1'o

. 1

. 2 . . 3 . 4. . 5. . 6

7

8

9

1'o

100 MILLISECOND INTERVALS

Fig. 6. Mean multiple unit discharge asssociated with placements located in either the centromedian or paracentral nuclei of the intralaminar nuclei of the thalamus, plotted as a function of 100-ms intervals after CS onset in z-score units normalized with respect to pre-CS (viz. baseline) discharge. A illustrates evoked unit discharge elicited from intralaminar nuclei in the paired CS/US group and B illustrates similar changes elicited from the group that received unpaired CS/US presentations. The data are shown (a) for each of the 4 blocks of 10 training trials on day 1 (ACQ) 1-4), (b) for the last 4 trials of pretraining and (c) for the last 10 trial-blocks of extinction.

extinction condition (largest P < 0.01). Post-tests comparing the M U A of the paired vs. unpaired conditions for each trial-block revealed that changes in the paired condition were greater than those of the unpaired condition on trial-block 2 (bins 4, 8, 9, and 10), on trial-block 3 (bins 3 and 4), and on trial-block 4 (bins 3, 4, 8 and 10). H o w e v e r , M U A discharge in the unpaired condition during trial-block 1 was also significantly greater than that o b t a i n e d in the paired condition on both trial-blocks 1 and 4, but not for any other trial-blocks. In summary, these data indicate that significant CSe v o k e d M U A increases occurred in M D in response to classical conditioning contingencies; these changes, further, f o r m e d a different pattern from that of animals that received u n p a i r e d CS/US presentations. The p e a k re-

sponse (viz. bins 3 and 4) increased over trial-blocks in the paired condition, but decreased systematically in the unpaired condition. M u l t i p l e u n i t activity in the i n t r a l a m i n a r n u c l e i d u r i n g classical c o n d i t i o n i n g

A comparison of C S - e v o k e d M U A in the p a i r e d and unpaired groups with intralaminar placements also revealed differences. A s can be seen in Fig. 6, t h e r e was a significant C S - e v o k e d increase in M U A , which occurred during the first three 100-ms bins in both p a i r e d and unpaired animals. H o w e v e r , unlike the M D placements, this increase was not systematically related to trial-block. A l t h o u g h increases in M U A in general s e e m e d to be somewhat g r e a t e r during acquisition training than during

TABLE I Pre-tone baseline multiple unit activity (HZ + 1 S. E. M.), for each of the training conditions studied, as a function o f electrode placement

Mediodorsal nucleus Intralaminar nuclei

Pretraining

A cq 1

A cq 2

A cq 3

A cq 4

A cq 5

A cq ~6

Ext 1

Ext 2

Ext 3

19.9 18.6

29.4 26.6

30.3 23.2

22.9 28.6

24.8 20.1

21.5 21.4

16.5 15.8

15.2 11.6

17.5 14.6

19.5 14.7

218 pretraining, A N O V A revealed that the changes during the initial 100-ms intervals were not significantly different in the paired and unpaired groups (P < 0.10). However, the changes in M U A during later 100-ms bins (e.g., 5-10) appeared to be greater in the paired than in the unpaired animals. There was a consistent decline from the 3rd to the 10th 100-ms bin in the unpaired animals, whereas CS-evoked M U A appeared to remain elevated over all ten 100-ms bins in the paired animals. In fact, Fig. 6 suggests that this increase in M U A during later bins was systematically related to acquisition trial-blocks. A N O V A again revealed significant bin (F9,~53 = 8.6, P < 0.0001) and trial-block (F5,666 = 9,8, P < 0.0001) main effects. The paired vs. unpaired condition × bin (F9,1~3 = 1.9, P < 0.05) and paired vs. unpaired condition × trial-block × bin (F45,666 = 1.7, P < 0.003) interactions were also significant. Post-tests revealed that on trialblocks 3 and 4, changes in bins 4-10 were significantly greater in the paired than in the unpaired animals (largest P < 0.01). This was not true of trial-blocks 1 and 2, however. Moreover, bins 4-10 of trial-blocks 3 and 4 were also significantly greater than those of pretraining in the paired but not the unpaired group (largest P < 0.01). However, bin 1 of extinction was significantly greater than bin 1 of all 4 acquisition trial-blocks. In summary, CS-evoked changes in M U A in the IL group, although somewhat different from that obtained in the MD group, also appeared to be associative in character, in that CS-evoked neuronal activity during later bins was significantly different from pretraining and extinction, and was systematically related to trial-blocks, in the paired, but not in the unpaired animals. Although not shown in Fig. 6, these differences persisted for the entire duration of the CS period. ANOVA of M U A in 100-ms bins during the last second prior to tone termination revealed a significant group effect (F1,18 = 6.16, P < 0.02), and trial effect (F4,~8 = 5.77, P < 0.0002). Post-tests revealed that, during the ten 100-ms bins just prior to CS offset, on trial-blocks 3 and 4, M U A remained elevated in the paired, but not in the unpaired group, compared to the pretraining and extinction conditions (largest P < 0.05). These results thus suggest that CS-evoked increases in IL M U A also had associative characteristics, but that the latency and duration of IL discharge was somewhat longer than that of MD and in fact persisted for the entire duration of the CS period.

Multiple unit activity: day 2 training CS-evoked changes in M U A activity during the 2nd session of associative training were similar to those obtained during the 1st session. Again there was an increase over blocks of 100-ms intervals for both the paired and unpaired MD and IL groups, but it was

considerably greater in the paired than in the unpaired conditions for both placements. This condition effect was significant in the MD group (F1,28 = 6.1, P < 0.02), but not intralaminar group (P > 0.10). However, for both groups, significant effects were found for trial-block (MD: F1,252 = 6.6, P < 0.01; IL: Fro62 = 7.2, P < 0.006), bin (MD: F9,252 = 7.3, P < 0.001; IL: Fro62 = 4.1, P < 0.001), and the paired vs. unpaired condition × bin interaction (MD: F9,252 = 2,0, P < 0.03; IL: F9,a6~ = 2.8, P < 0.004). Moreover, in bins 5-10 M U A was significantly greater in the paired than in the unpaired animals in the IL (P < 0.01), but not the MD (P > 0.10) group.

Multiple unit activity: extinction Tone-evoked M U A during the last 10 trials of extinction was discussed above in comparison with animals that received paired vs. unpaired CS/US training. CS-evoked increases in M U A in the animals with both MD and IL placements were considerably smaller during extinction than during either session of classical conditioning. Separate ANOVAs of M U A for the IL and MD groups revealed that neither the condition nor condition × post-CS bin effects was significant for either group during extinction (all P's > 0.10). Pre-tone heart rate and MUA The effects of the training schedules on maintained M U A is shown in Table I. Maintained (pre-CS) M U A appeared to increase as a result of initial training in both the IL and MD groups. An A N O V A of this data revealed a significant effect over training conditions, for both placements and conditions. Post-tests revealed that maintained M U A on the 1st day of conditioning was significantly greater than that obtained during pretraining in all animals (P < 0.01). However, no other differences were significant (all P's > 0.10). In general, baseline (viz., pre-tone) H R was somewhat slower in the unpaired than paired groups with both the IL and MD placements, but these effects were not statistically significant for any stage of training. Baseline H R (BPM) ranged from 215.1 to 231.5 in the paired condition and from 199.6 to 211.4 in the unpaired condition during various stages of training. A statistically significant trials effect was obtained for both the MD and IL analyses during all 3 stages of training (largest, P < 0.05), reflecting a decline in baseline H R over trials in all animals. However, no other significant baseline H R effects were obtained. Thus, baseline differences in H R were unremarkable in the present study. Relationships between CS-evoked MUA and conditioned bradycardia The present results suggest that associative training-

219 induced changes in CS-evoked MUA, which paralleled the acquisition of the H R CR, occurred in both the IL and MD nuclei of the thalamus in the rabbit. However, an important question concerns which aspects of this evoked discharge, if any, can be shown to be specifically related to the H R responses to the CS during different phases of training. Accordingly, the neuronal and behavioral data were submitted to correlational analyses using the Pearson Product Moment correlation coefficient as the test statistic. No significant correlations were found between M U A and heart rate changes in response to the tone during pretraining in either the MD or IL groups. Analysis of absolute M U A and H R response levels for each animal averaged over the four 10-trial blocks of classical conditioning, also indicated no apparent relationship between the two, since none of these correlation coefficients were significant. However, since the initial facilitatory effects of conditioning on M U A occurred within the first 300 ms after tone onset, the relationship of this unit response component to conditioned bradycardia was evaluated by comparing the largest magnitude H R change, during post-CS iBI 10, with M U A Z-scores occurring in the 2nd and 3rd 100-ms bins for both the MD and IL groups. These correlation coefficients were computed separately for the pretraining, acquisition, and extinction phases of the study. None of these correlation coefficients were significant for either the paired or unpaired conditions in the IL groups. However, animals with MD placements that received paired training showed relatively high positive correlations between CS-evoked M U A and the heart period associated with the 10th IBI (r = 0.47, P < 0.07 and r = 0.59, P < 0.02 for the 2nd and 3rd bins, respectively) for the first acquisition session. These correlations were obtained only in the paired MD group, and only during the first acquisition session; no other correlation coefficients were significant for either the paired or the unpaired groups with MD placements. Since the largest changes for the IL-paired groups occurred during the 10th 100-ms bin following CS onset, M U A occurring during this interval was also correlated with the 10th IBI for both placement groups. However, no systematic pattern of significant correlations was observed in either the IL or MD groups as a result of this analysis. DISCUSSION The results of these experiments indicate that traininginduced enhancement of CS-evoked discharge occurs in both the MD and IL nuclei of the dorsal thalamus during classical conditioning training. Moreover, the development of classically conditioned bradycardia was related to

the increase in CS-evoked M U A in MD, but not in the IL nuclei, early in training. It is, however, unknown whether this change in MD neuronal activity in fact contributes importantly to the development or the expression of the H R CR. Previous studies from our laboratory demonstrated that although deficits in the parasympathetically-mediated conditioned bradycardiac CR were produced by medial PFC lesions 3'4, both parasagittal thalamic knife cuts, which sever PFC-MD connections, and ibotenic acid lesions of MD, have relatively little effect on either acquisition or expression of conditioned bradycardia in the rabbit 2~5"6. These studies thus suggest that MD neural activity is not necessary for either acquisition or expression of the bradycardiac CR. Other studies have suggested, however, that damage to prefrontal-MD connections produces a dramatic decrease in the sympathetic component of the cardiovascular changes that accompany classical eyeblink conditioning 5'9. These findings have led us to suggest 5 that, whereas the PFC is involved in the initial processing of stimuli for informational value, MD is involved in the selection of somatomotor responses to deal effectively with environmental contingencies, as originally suggested by Vanderwolf 32. Such a process might be expected to engage the necessary sympathetic mechanisms for elaborating an appropriate somatomotor response. In support of this hypothesis, MD lesions that attenuated the sympathetic mechanisms normally associated with somatomotor conditioning also impaired acquisition of the classically conditioned somatomotor eyeblink response 5. The finding in the present study that MD M U A in the unpaired group was greatly augmented during early training, when the animals were first exposed to unpaired tones and eyeshocks (viz., during trial-block 1 of acquisition training), and the concomitant occurrence of tone-evoked tachycardia in these animals, is also compatible with the association of MD with sympathetic function. It is also of some interest that the IL nuclei of the thalamus showed CS-evoked associative changes as a result of CS/US pairings during classical conditioning, although the pattern of M U A increases in these nuclei was considerably different from those obtained with MD placements. The peak response of the IL cells was considerably later than that of MD cells. Also, the evoked response in the IL, while apparently related to information processing, was unrelated to the elaboration of the conditioned H R response. A major question then concerns the functional significance of the CS-evoked increases in the IL nuclei. As noted above, although both IL and MD project to the prefrontal region, a very strong projection from IL to the neostriatum has been demonstrated in a variety of species 8'a6'28. The neostriatal

220 region, as well as other extrapyramidal structures, are known to be involved in the elaboration of classically conditioned somatomotor responses 18'24"29'3°. It is thus possible that IL is also involved in somatomotor response selection, as we have previously hypothesized for MD 2. For the most part, the results reported here are compatible with earlier studies in rat and rabbit. Orona and Gabriel 21"22, for example, reported CS-evoked differential short latency increases in MUA in both the PFC and MD in rabbits during a CS+ tone signalling foot shock during an instrumental running wheel avoidance response. Such increases occurred during both pretraining and conditioning sessions and were significantly greater than changes recorded during a similar tone that never signaled foot shock. In the PFC, this response showed its peak between 100 and 200 ms, and, in MD, between 200 and 300 ms, which corresponds well with the latencies reported here and in our previous study of medial PFC M U A 13. Orona and Gabriel 21"22, however, recorded CS-evoked neuronal changes in these areas over much longer periods of time than reported here, since operant avoidance requires many trials before the response is learned; animals were also required to learn the running wheel response to reversal, as well as acquisition criteria. Nevertheless, the data of the present study are compatible with most aspects of those reported by Orona and Gabrie121'22; however, see below for a major exception. An earlier study in the rat reported by Disterhoft and Olds 1° showed significant differential MUA increases in a variety of thalamic and cortical structures in response to a tone CS+ that predicted food, compared to a second tone CS- that was never paired with food. Many more trials were also utilized in the Disterhoft and Olds 1° study than in the present experiments, since, again a somatomotor response (i.e., increased general activity) was the criterion behavioral response. These investigators reported increased MUA elicited from MD during a CS+ tone that was greater than that recorded during CSpresentation, although the magnitude of this increase was considerably greater in other thalamic and cortical areas than in MD. For example, the magnitude of this increase was considerably greater in IL than in MD in the Disterhoft and Olds ~° study. The effects of conditioning on CS-evoked MUA in the present study appeared to be manifested throughout all of MD, including both medial (n = 6) and lateral (n = 13) placements. However, Orona and Gabriel z~ reported that MUA in medial MD was not differentially related to the reinforced CS+ and non-reinforced CS-, whereas that in lateral MD was. It is impossible to resolve these differences at the present time. We have too few medial MD placements for statistical analysis and there were

many procedural differences between the present study and those of Orona and Gabriel z1'22, as described above. As can be observed in Fig. 1, electrode placements were also found in both the centromedian and paracentral nuclei of IL. Again, no differences were noted between placements in these two different intralaminar nuclei, although, as with MD, a larger sampling of cells in these respective areas of IL will be necessary to draw firm conclusions in this regard. Taken as a whole, the present results, along with earlier studies 1°'21~22, suggest that significant changes in discriminative unit activity may occur in a variety of neocortical and thalamic structures as a result of conditioning tasks. However, the present study suggests that the occurrence of a learned specific somatomotor response is not required for this increase in neuronal discharge to occur. The significant correlations obtained in our previous study between PFC MUA a3, and the correlations in the present study between M U A in MD and conditioned HR changes during the initial trials of training, suggest that these changes may be importantly involved in the elicitation of the autonomic changes that precede, and later accompany, learned somatomotor behaviors. These data not only provide additional evidence that the thalamic-PFC axis is involved in the information processing necessary for associative learning to occur, but they further suggest that the specific somatomotor responses typically employed to index such learning may be unrelated to such early occurring neuronal events. On the other hand, there must be a CNS interface between structures involved in information processing and those involved in generating the adaptive behaviors that ultimately occur as a result of such processing. The thalamic-PFC axis may provide such an interface by participating in a somatomotor response selection process. It is important to note, however, that the acquisition, shaping, and storage of the motor programs underlying such skeletal responses is almost certainly associated with neither the PFC nor its thalamic connections 3"5'6. Current evidence suggests that extrapyramidal motor structures may provide a substrate for this kind of somatomotor plasticity 18'24'z9"3°. Based on this model, interference with learned somatomotor responses by MD damage, as previously reported 17"2°'33, would be due to MD's participation in a response selection process, rather than in the acquisition of the somatomotor response per se. In summary, the present data suggest that associative changes occur in MD and IL when rabbits are exposed to classical conditioning contingencies. These changes occurred relatively early during acquisition, prior to the development of somatomotor conditioning, as demonstrated in previous experiments, and with parameters that

221 i n d e e d a r e n o t o p t i m a l f o r s o m a t o m o t o r c o n d i t i o n i n g 23. U n c l e a r at t h e p r e s e n t t i m e , h o w e v e r , is t h e e x t e n t to w h i c h t h e s e C S - e v o k e d c h a n g e s in n e u r o n a l activity a r e d i r e c t l y r e l a t e d t o t h e c o n c o m i t a n t l y a s s e s s e d classically conditioned bradycardiac responses.

REFERENCES 1 Benjamin, R.M., Jackson, J.C. and Golden, G.T., Cortical projections of the thalamic mediodorsal nucleus in the rabbit, Brain Research, 141 (1978) 251-265. 2 Buchanan, S.L., Mediodorsal thalamic lesions impair differential Pavlovian heart rate conditioning, Exp. Brain Res., 73 (1988) 320-328. 3 Buchanan, S.L. and Powell, D.A., Cingulate cortex: its role in Pavlovian conditioning, J. Comp. Physiol. Psychol., 96 (1982) 755-774. 4 Buchanan, S.L. and Powell, D.A., Cingulate damage attentuates conditioned bradycardia, Neurosci. Lett., 29 (1982) 261268. 5 Buchanan, S.L. and Powell, D.A., Parasagittal thalamic knife cuts retard Pavlovian eyeblink conditioning and abolish the tachycardiac component of the heart rate conditioned response, Brain Res. Bull., 21 (1988) 723-729. 6 Buchanan, S.L. and Powell, D.A., Parasagittal thalamic knife cuts and cardiac changes, Behav. Brain Res., 32 (1989) 241-253. 7 Buchanan, S.L., Valentine, J.D. and Powell, D.A., Autonomic responses are elicited from medial but not lateral frontal cortex in rabbits, Behav. Brain Res., 18 (1985) 51-62. 8 Buchanan, S.L., Powell, D.A. and Thompson, R.H., Prefrontal projections to the medial nuclei of the dorsal thalamus in the rabbit, Neurosci. Lett., 106 (1989) 55-59. 9 Buchanan, S.L., Thompson, R.H. and Powell, D.A., Midline thalamic lesions enhance conditioned bradycardia and the cardiac orienting reflex in rabbits, Psychobiology, 17 (1989) 300-306. 10 Disterhoft, J.E and Olds, J., Differential development of conditioned unit changes in thalamus and cortex of rat, J. Neurophysiol., 35 (1972) 665-679. 11 Domesick, V.B., Projections from the cingulate cortex in the rat, Brain Research, 12 (1969) 296-320. 12 Faull, R.L.M. and Mehler, W.R., Thalamus. In G. Paxinos (Ed.), The Rat Nervous System, Academic Press, New York, NY, 1985, pp. 129-168. 13 Gibbs, C.M. and Powell, D.A., Neuronal correlates of classically conditioned bradycardia in the rabbit: studies of the medial prefrontal cortex, Brain Research, 442 (1988) 86-96. 14 Gibbs, C.M., Prescott, L. and Powell, D.A., A comparison of CS-evoked multiple unit activity in medial and insular prefrontal cortex during classical heart rate conditioning in the rabbit, Behav. Brain Res., submitted. 15 Groenewegen, H., Organization of the afferent connections of the mediodorsal thalamic nucleus in the rat, related to the mediodorsal-prefrontal topography, Neuroscience, 24 (1988) 379-431. 16 Herkenham, M., Laminar organization of thalamic projections to the rat neocortex, Science, 207 (1980) 532-535. 17 Isserhoff, A., Rosvold, H.E., Galkin, T.W. and GoldmanRakic, P.S., Spatial memory impairments following damage to the mediodorsal nucleus of the thalamus in rhesus monkeys, Brain Research, 232 (1982) 97-113.

Acknowledgements. This research was supported by VA Institutional Research Funds awarded to the Wm. Jennings Bryan Dorn VA Medical Center. We thank Charles Gibbs for his helpful suggestions during the conduct of this study. We also thank Fred Fenimore for the development of the software used, Elizabeth Hamel for her expert secretarial assistance, and the University of South Carolina Medical Illustration Department for preparing the illustrations.

18 Kao, K.-T. and Powell, D.A., Lesions of the substantia nigra retard Pavlovian eyeblink but not heart rate conditioning in the rabbit, Behav. Neurosci., 102 (1988) 515-525. 19 Kazis, E., Milligan, W.L. and Powell, D.A., Autonomic-somatic relationships: blockade of heart rate and corneoretinal potential, J. Comp. Physiol. Psychol., 84 (1973) 98-110. 20 Markowitsch, H.J., Thalamic mediodorsal nucleus and memory: a critical evaluation of studies in animals and man, Neurosci. Biobehav. Rev., 6 (1982) 351-380. 21 Orona, E. and Gabriel, M., Multiple-unit activity of the prefrontal cortex and mediodorsal thalamic nucleus during acquisition of discriminative avoidance behavior in rabbits, Brain Research, 263 (1983) 295-312. 22 Orona, E. and Gabriel, M., Multiple-unit activity of the prefrontal cortex and mediodorsal thalamic nucleus during reversal learning of discriminative avoidance behavior in rabbits, Brain Research, 263 (1983) 313-329. 23 Powell, D.A. and Kazis, E., Blood pressure and heart rate changes accompanying classical eyeblink conditioning in the rabbit (Oryctolagus cuniculus), Psychophysiology, 13 (1976) 441-447. 24 Powell, D.A., Mankowski, D. and Buchanan, S.L., Concomitant heart rate and corneoretinal potential conditioning in the rabbit (Oryctolagus cuniculus): effects of caudate lesions, Physiol. Behav., 20 (1978) 143-150. 25 Powell, D.A., Buchanan, S.L. and Hernandez, L.L., Electrical stimulation of insular cortex elicits cardiac inhibition but insular lesions do not abolish conditional bradycardia in rabbits, Behav. Brain Res., 17 (1985) 125-144. 26 Rose, J. and Woolsey, C., The orbitofrontal cortex and its connections with the mediodorsal nucleus in rabbit, sheep and cat, Assoc. Res. Nervous Mental Dis., 27 (1948) 210-232. 27 Rosene, D.L., Roy, N.J. and Davis, B.J., A cryoprotection method that facilitates cutting frozen sections of whole monkey brains for histological and histochemical processing without freezing artifact, J. Histochem. Cytochem., 34 (1986) 1301-1315. 28 Scheibel, M.E. and Scheibel, A.B., Input-output relations of the thalamic nonspecific system, Brain Behav. Evol.~ 6 (1967) 332-358. 29 Thompson, R.E, The neurobiology of learning and memory, Science, 233 (1986) 941-947. 30 Thompson, R.E, McCormick, D.A., Lavond, D.G. Clark, G.A., Kettner, R.E. and Mauk, M.D., The engram found? Initial localization of the memory trace for a basic form of associative learning. In J.L. Sprague and A.N. Epstein (Eds.), Progress in Psychobiology and Physiological Psychology, Academic Press, New York, NY, 1983, pp. 167-196. 31 Urban, U. and Richard, P., A Stereotaxic Atlas of the New Zealand Rabbit, Charles C. Thomas, Springfield, IL, 1972. 32 Vanderwolf, C.H., Limbic-diencephalic mechanisms of voluntary movement, Psychol. Rev., 78 (1971) 83-113. 33 Zola-Morgan, S. and Squire, L.R., Amnesia in monkeys after lesions of the mediodorsal nucleus of the thalamus, Ann. Neurol., 17 (1985) 558-564.