Beha vioural Brain Research, 25 ( 1987) 117 - 129 1 17 Elsevier

BBR 00692

Role of dorsomedial thalamic nucleus and piriform cortex in processing olfactory information

U. Staubli, F. Schottler and D. Nejat-Bina Center for the Neurobiology of Learning and Memory, University of California, Irvine, CA 92717 (U. S.A.)

(Received 31 July 1986) (Revised version received 28 April 1987)

(Accepted 28 April 1987)

Key words: Olfactory discrimination learning; Higher mammalian memory; Dorsomedial thalamic nucleus; Piriform cortex; Procedural impairment; Encoding deficit; Anterograde amnesia; Rat

Four experiments were conducted to characterize the role of primary and secondary olfactory projection areas (piriform cortex and dorsomedial thalamic nucleus (DMN)) in olfactory information processing. Rats had to learn to discriminate between odors that were simultaneously released from different arms of an automated olfactory maze. When standard training conditions were used, damage of the DMN severely impaired both preoperatively trained and naive animals in acquiring an odor discrimination set (i.e. in most problems no learning was demonstrated). An additional group of DMN animals that received 4 times the standard amount of daily trials was unable to acquire the first two problems but successfully solved the third and all subsequent discriminations. Analysis of performance patterns suggested that destruction of the DMN initially leads to a strong procedural impairment that can be overcome by extensive training. After solving the third problem the animals with DMN damage required much less training to reach the learning criterion but generally made more errors than controls. Transfer of savings rarely occurred when a problem was repeated. Whether this secondary learning deficit observed in later discriminations is due to a specific effect of the lesion on the encoding of olfactory cues and thus on memory formation, or due to a disturbance in the regulation of emotional factors such as motivation, arousal, and attention is discussed. Lesions of the thalamus that spared the DMN had no effect on learning or retention of olfactory discriminations. Animals with ablations of the piriform cortex only acquired odor discriminations if they had been trained in the olfactory maze before the lesion. Moreover, their performance depended on the odor quality: they had great difficulty learning complex cues consisting of several odorants and learned simple odors virtually identical to control rats. The results indicate that an intact piriform cortex is needed to acquire the procedures involved to perform an olfactory discrimination task as well as to build neural representations of olfactory cues.

INTRODUCTION

Rodents are among the most widely used animals in neurobiological studies of learning and memory. Even though popular belief ascribes excellent smelling capacities to the rat, ironically only very few learning studies have employed olfactory cues. This lack of interest in olfaction is also surprising in view of recent behavioral and anatomical evidence suggesting that the olfactory

modality in rats is well suited to characterize memory mechanisms typically found in 'higher' mammals. The nature of rat olfactory learning is qualitatively very different from learning that requires use of the auditory, visual or other sensory modalities. Rats confronted with succes- sive olfactory discriminations or reversals demon- strate 'learning set' formation 9"~ 9, a learning mode characteristic for 'higher' mammals 4. The term learning set is used by animal psychologists to

Correspondence: U. Staubli, Center for the Neurobiology of Learning and Memory, University of California, Irvine, CA 92717, U.S.A.

0166-4328/87/$03.50 © 1987 Elsevier Science Publishers B.V. (Biomedical Division)

118

describe interproblem learning occurring during a series of discriminations having the same method of solution. Regardless of the various interpreta- tions raised, such as 'rule learning' or 'learning to learn' to account for this phenomenon, this type of learning is commonly regarded as evidence for higher order learning; from an anatomical point of view, the olfactory system is special in that its primary structures, the piriform and the entorhinal cortex, are cortical in nature. They are innervated by the lateral olfactory tract (LOT) and in turn connect directly (via one synapse) with the hippocampus and the dorsomedial thalamic nucleus (DMN), two subcortical struc- tures that seem particularly relevant to human memory (e.g. refs. 8, 20). Thus, a major part of the anatomical substrate of olfaction in rats is identi- cal with brain areas that appear to be critically involved in memory processing in humans; damage to these structures causes severe anterograde amnesia.

In rats, lesions of the hippocampus consistently result in an inability of rats to solve spatial mazes, a finding which led to the formulation of several hypotheses on hippocampal function in rodents. Among these, the two most influential, though still controversial ones, propose that the hippocampus is responsible for: (1)building 'spatial maps' of the environment"~; and (2) the formation of work- ing memory ~1. Neither of the two hypotheses can fully account for the hippocampal deficits found in humans with temporal lobe damage. This raises the question of whether phylogenetic differences between man and rodents, or a limited set of experimental data on the rodents' part account for the observed discrepancies.

Much less is known about the role of the rat DMN in learning. A number of studies show a deleterious effect of large DMN lesions on avoidance learning (e.g. ref. 12) as well as on discrimination learning (e.g. ref. 2), but they all lack the demonstration of a correlation between the observed deficits and damage restricted to the DMN. Recent data from studies on rodents sug- gest a direct involvement of the DMN in olfactory learning: rats with DMN lesions showed impaired acquisition of olfactory discriminations i.~ s. The deficits observed were selectively related to

destruction of the DMN. In addition, the severity of the impairment appeared to depend on qualita- tive aspects of the task such as degree of difficulty and novelty of a given olfactory discrimination. All things considered, the data on the role of the DMN in learning are very limited and may be attributable to specific training and testing condi- tions.

It has recently been demonstrated that rats with hippocampal dysfunction show rapid forgetting of olfactory information 22"23, just as humans with temporal lobe damage do for most kinds of newly acquired data 2°. Thus, olfactory memory in the rat might be an ideal model system to study and characterize hippocampal involvement in human memory. The possibility that such a formal homology could similarly be established for the DMN seems reasonable considering there is already evidence of an involvement of this nucleus in rat olfactory learning 1. J 8.

The olfactory projections to both D M N 6"14 and hippocampus 3,5 originate in the piriform cortex, the primary olfactory projection site. Little be- havioral data is available on the role of the piriform cortex itself in olfaction. The results of two studies suggest that LOT projections rostral to the amygdala, i.e. the anterior piriform cortex and the olfactory tubercle 16"17 are sufficient to support odor discrimination; transection of the LOT at this level does not cause anosmia ~7. That the piriform cortex might play a critical role in olfactory learning is indicated by recent evidence of a long-lasting potentiation of synaptic re- sponses in the bulbar-piriform projections elicited by behaviorally meaningful electrical stimulation of the LOT 1~'24. Here we describe an experiment in which the effect of partial lesions of the piriform cortex was tested on learning of a series of olfactory discriminations.

MATERIALS AND METHODS

Subjects Subjects were adult male rats (Long Evans)

that weighed 300-350 g and were housed indi- vidually. Behavioral training and testing took place during the dark phase of their reversed light-dark cycle. The animals were kept on a

119

deprivation schedule allowing 7 min access to water a day while food was available ad libitum.

Lesions Three types of lesions were performed. One

was aimed at the D M N (DMN group, n = 15) one spared the D M N and destroyed the surrounding thalamic nuclei (thalamic control group, n = 9), and a third one was aimed at the anterior piriform cortex (piriform group, n --- 8). Before surgery the rats were anesthetized with i.p. injections of sodium pentobarbital (Nembutal 50mg/kg). Bilateral electrolytic lesions were placed under stereotaxic guidance with the toothbar set at - 3 . 3 mm (DMN: AP -2 .3 , L 1.5, DV 5.4 and AP - 3.3, L 1.5, DV 5.65 at an angle of 10.6 ° and 9 ° from the vertical midline, respectively. Thalamic control: AP - 1.3, L 2.2, DV 5.6 and AP - 4.3, L 2.2, DV 5.5 at an angle of 15.5 ° from the vertical midline. Piriform: AP 3.2, L 1.0, DV 6.75 and AP 3.2, L 2.0, DV 6.6 as well as AP 2.2, L 1.0, DV 8.0 at an angle of 5 °, i0 ° and 21 ° from the vertical midline, respec- tively). Current of 1 mA was passed during 45 s at each placement for the thalamic lesions and during 60 s at each placement for the piriform cortex lesions. For sham lesions the electrode was lowered to the same location of the D M N (n = 8) and piriform cortex (n = 7), respectively, but no current was delivered. Olfactory training began 10 days after surgery.

Histology At the end of the experiments, all animals that

had received a lesion were deeply anesthetized with Nembutal and perfused transcardially with physiological saline followed by 10~o form- aldehyde solution (in 0.1 M phosphate buffer). The brains were removed and stored in the form- aldehyde solution for several days. The day before sectioning they were put in 20 ~o saccharose solu- tion (in 0.1 M phosphate buffer) overnight. The brains were then frozen, sectioned at 45/~m and every third section was stained with Cresyl violet for histological examination of the lesions. The sections of all brains were examined microscopi- cally and were reconstructed using a standard stereotaxic atlas ~3. Representative lesions are schematically illustrated in Figs. 5-7.

Apparatus The original demonstration of learning set per-

formance for olfactory cues was conducted in an operant chamber, using a conventional bar- pressing paradigm ~9. We designed an olfactory maze consisting of different arms in which the rat was required to locate the odor source and then move toward it to find a reward, a sequence an animal is likely to perform in its natural habitat. The olfactory apparatus consisted of 4 covered alleys made out of dark plexiglass (52 cm long, 10 cm wide, 12 cm high) that extended from a rectangular center platform (26 x 26 x 18 cm) and ended each in a goal box (18 x 19 x 18 cm).

Before each trial the 3 different odors were randomly assigned to 3 arms of the maze while the unassigned 4th arm served as starting alley. By blocking one arm permanently the apparatus could be converted into a 3-arm maze with the discrimination problems consisting of two odors.

Odors Odors were created by forcing a clear airstream

at a fixed pressure (flow meter) through l-liter flasks containing different, commercially availa- ble synthetic odors and extracts (Imitation Flavors and Extracts, McCormick, Inc.) diluted in 500 ml of water. The outlet of each flask was connected to a tube (tygon tubing i.d. 3/16 inch) that carried the odorized air and ejected it into the maze via openings in the goal boxes. To insure clearing of odors from the maze an exhaust fan was built into the ceiling of the center platform to pull the odors down the arms and out of the testing area.

Concentrates were diluted individually such that they were perceived by humans as being of similar intensity. It has been observed that strength of the cues is not a critical variable in tasks involving different odors (Eichenbaum, personal communication). We found that rats are capable of learning two identical smells that differ in intensity (1 : 2/3), but that the animals need 2 to 3 times more trials than when learning two dif- ferent odors (unpublished data). Together, these findings suggest that rats prefer to use a difference in quality over a difference in concentration as a means to discriminate between two odors.

120

In an attempt to increase the difficulty of the discriminations we used odors that were com- bined from several chemicals (complex odors), with one chemical being different between two odors of a given pair (e.g. ab + vs abc - ). Well- trained rats easily learn to discriminate such odors 2~. A stock of 200 synthetic aroma chemi- cals were available (supplied by International Flavors and Fragrances, New York). The con- struction of complex odors requires elements of equivalent strength since dominant elements integrated with lesser components might suppress these components. In order to be able to evaluate the strength of individual odors, each aroma chemical was diluted several times in steps of 10 to a concentration indiscriminable from water. The amount of each elementary odor to be added to form the complex cue was determined from this dilution scale.

To avoid contamination of odors by residual traces of previously used odors the tubing was always replaced between individual sessions.

Procedure Operation of the maze was fully controlled by

computer (Commodore 64) that randomly assigned the odors to the arms at the beginning of each trial. The data were monitored on screen and printed out after each session. At the beginning of the first trial the animal was set into the goal box that had not been assigned an odor. By moving down the arm, the animal arrived at the center platform where it had to decide, on the basis of the odors that were released, which arm to follow. A waterspout was hidden in the floor at the far end of each goal box. The odor was ejected through one of 3 openings in the wall behind the spout, and a photosensitive cell was mounted across the walls 5 cm in front of the spout. Detection of the animal's presence automatically shut the door to the goal box enclosing the rat, and depending on its choice, 5 drops of water were released from the floor, or else a small light bulb was turned on for 10 s to signal a wrong response. The odors turned off 10 s after each choice and were randomly reassigned and released into the 3 empty goal boxes. Sixty seconds later, the door of the goal box, which was now the start box, opened and the

animal was free to move to the center platform and choose one of the odors again.

Training Before being trained the animals were water

deprived for 48 h. They were then "allowed to explore the maze for 30-45 min on two consecu- tive days. No odors were present during these familiarization sessions, and water was delivered each time the animal entered a goal box.

Two training schedules were used: (1) a series of different discriminations between 3 odors was presented on consecutive days during 25 trials each (Expts. 1 and 2); and (2) a series of dis- criminations between two odors was presented on consecutive days (100 daily trials) or until a criterion of 8 correct choices in 10 consecutive trials was met. Some of the problems were repeated on different days (Expts. 3 and 4).

In Expts. 1 and 3 the lesions were done in naive animals, in Expts. 2 and 4 the animals were well trained before the lesion.

Statistics A two-way analysis of variance for repeated

measures was used for primary data analysis in all experiments, and post-hoc comparisons were done using the Scheffe test.

RESULTS

Experiment 1 Three groups of rats were tested, two with

thalamic lesions (DMN, n = 6; thalamic control, n = 5) and one with sham lesions (sham, n = 4).

Fig. 1 shows postlesion performance of all groups in a series of 8 discrimination problems. Comparison of group performances were based on a criterion of 8 correct choices in 10 consecu- tive trials and indicated that DMN lesions did affect the ability to acquire an odor discrimination set (ANOVA: F = 54.386, df = 2/96, p < 0.000). Post-hoc Scheffe comparisons revealed that the scores of the DMN group were consistently and significantly different from the scores of the sham lesion group starting with discrimination 2 (prob- lems 2, 3, 4, 5, 6, 7 and 8, p < 0.01) 2, 3, 4, 5, 6, 7 and 8 as well as from the scores of the thalamic

121

2 7

23 - - -

21 "x,, \T /

z- 17 o SHAM

l \ \ T 1 (.~ 15 o THAL CONTROL

7

5

B

1 I I ~ l l I l 13 I 2 4 5 6 7

Discrirninotions

Fig. 1. Performance of control and brain-damaged rats in a series of discrimination problems with different triads of odors. A lesion of the D M N had been placed in 6 rats (DMN) , a thalamic lesion excluding the D M N in 5 rats (THAL. CONTROL) and a sham lesion in 4 additional rats (SHAM). Twenty-five trials per problem were given. Group performances were calculated as mean number of trials to reach a criterion of 8 correct choices in 10 consecutive runs

(mean + S.E.M.).

control group (problems 3, 4, 5, 6 and 8, p < 0.01 ; problem 7, p < 0.05). In the second discrimina- tion sham control animals performed significantly better than the thalamic control group (p < 0.01), but in all the following problems the scores of the two control groups were no longer statistically different.

Experiment 2 This experiment was identical with the preced-

ing one except that the animals received olfactory training (9 problems) before the lesion. Three groups of rats were tested, two with a thalamic lesion (DMN, n = 5 ; thalamic control, n = 4) and one group with a sham lesion (sham, n = 4). Six of the 8 discriminations presented postoperatively were novel; problems 3 and 8 had been acquired preoperatively as problems 6 and 5.

As shown in Fig. 2A, performance levels of all groups were comparable after acquisition of 9 consecutive problems (ANOVA: F = 1.503, df = 2/81, p < 0.273).

In contrast, as Fig. 2B shows, thalamic lesions dramatically affected the ability to acquire olfactory discriminations (ANOVA: F = 16.930, dF = 2/67, p < 0.001). Post-hoc Scheffe tests revealed a significant impairment of the D M N

25

25

21 19

A 0 SHAM

• , , \ / - X T T

' \ ~ / X

1 I I J i i I i

I 5 4 5 6 7 8 9

DiscrlminQtions

17

15

13

II

9 7 5 5 I

25 23 - Q~ II . . . . IIx X ~t t-~-" j1 . i ! !!!!~ 21 - "~----I ~ 19 -

=o 17 - ~ ", 15- CONTROL

o 15- \

~ 9

7

5

3

I - ~ ~

I 2 3 4 5 6 7 8 Discriminations

Fig. 2. Preoperative (A) and postoperative (B) performance of control (SHAM, n = 4) and brain-damaged rats (DMN, n = 5; THAL. CONTROL, n = 4) in a series of discrimina- tion problems with different triads of odors. For each problem the mean number of trials per group to reach a criterion of 8 correct choices in 10 consecutive runs is illustrated (mean + S.E.M.). Twenty-five trials per problem

were given.

group compared to sham control animals in all discriminations except the second (problems 1, 3, 4, 6 and 7, p < 0.01 ; problems 5 and 8, p < 0.05). No improvement was obvious when the two familiar preoperatively acquired problems were presented. The thalamic control group showed a noticeable learning deficit in the first 3 problems. Since they mastered all the remaining problems in a manner comparable to control animals, their initial deficit probably reflected an unspecific effect of the lesion. The scores of these two control groups did not differ significantly except in problems 1 (p < 0.01) and 3 (p < 0.05).

Compared to performance of the thalamic control group D M N animals learned significantly

122

slower in problems 3, 4 and 6 (p < 0.01) and in problems 5 and 7 (p < 0.05).

Experiment 3 Three groups of rats were tested, one with a

thalamic lesion (DMN, n = 4), one with damage of the piriform cortex (piriform, n = 4) and one group with a sham lesion of the piriform cortex (sham, n = 4). Prompted by the severe learning impairment of DMN animals observed in the preceding experiments, we attempted to decrease the complexity of the task by using the 3-arm version of the olfactory maze with the discrimina- tions consisting of two odors. In addition, the problems were presented for 100 trials each or until the learning criterion was met, and problems 2, 3, 4, 5 and 6 were repeated on consecutive days. Furthermore, the first 4 problems were identical with the last 4, i.e. 1 = 10, 2 = 9, 3 = 8, 4 = 7).

Fig. 3 shows performance of the 3 groups in a series of odor discriminations. Controls animals learned as expected: after using 27 trials in the first discrimination to reach criterion, they acquired the next 5 problems in 5 trials and the following 4 problems in one or two trials.

Comparison of group performances indicated that the lesions drastically affected the ability to solve olfactory discriminations (ANOVA:

OSHAM

OPIRIFORM

mDMN

40 / / "

20 ", . """

I 2 3 4 5 6 7 8 9 I0

D iscr imlno l ions

Fig. 3. Performance of control and brain-damaged rats in discrimination problems with successive pairs of odors. A lesion of the piriform cortex had been placed in 4 rats (PIRIFORM), a lesion of the DMN in 4 rats (DMN) and a sham lesion in 4 rats (SHAM). Daily training was conducted for a maximum of 100 trials or until 8 correct choices in 10 consecutive runs were made. Five problems (2, 3, 4, 5, 6) were repeated on consecutive days. The first 4 problems were identical with the last 4. Group scores represent mean trials

to reach criterion (mean _+ S.E.M.).

F = 60.674, df = 2/90, p < 0.000). DMN animals showed no sign of acquiring the first two prob- lems, but they clearly improved during the follow- ing discriminations and were able to meet the learning criterion. Their scores were comparable to control animals solving the very first problem. However, reaching the learning criterion usually did not result in better performance when the same odors were presented again on the following day. Post-hoc Scheffe analysis revealed that the D M N group was significantly impaired compared to sham-lesioned animals in problems 1, 2, 4 (day 2), 5 (day 1) and 6 (day 1) with p < 0.01 as well as in problems 3 (day 2), 5 (day 2) and 7 with p < 0.05.

A partial ablation of the piriform cortex had a more severe impact on olfactory discrimination learning than D M N lesions. On most occasions the animals never acquired the problems, and repeated presentation of the same two odors on successive days had no influence on learning. Post-hoc Scheffe analysis revealed that the scores of the piriform group were significantly different from the scores of sham controls (p < 0.01) on all days and discriminations. They were also signifi- cantly impaired (p < 0.01) relative to D M N ani- mals in problems 4 (both days), 6 (both days), 8 and 10, and with p < 0.05 in problems 3 (day 2) and 5 (day 1). Performance of two animals was consistently superior: 3 discriminations out of 10 were acquired by all subjects, but only two ani- mals were able to learn 5 problems.

Experiment 4 Fig. 4 shows performance of animals with piri-

form cortex lesions (n = 4) and sham lesions (n = 3) before (top) and after (bottom) surgery. Group performances before the lesion were indistinguishable from each other (ANOVA: F = 1.067, df = 1/76, p < 0.349) but differed sig- nificantly after the lesion (ANOVA: F-- 21.888, df = 1/82, p < 0.005).

The initial drastic learning deficit of the piriform group decreased steadily across the first few problems. In the 4th discrimination performance of the two groups was virtually identical. There- fore, to test the idea that the discrimination of complex as opposed to simple smells might be

123

= 90 ,3

80

70 o

60

5O

40

30

2O

I 2 3 4 5 6 7 8 9

OSHAM I OPtRIFORM

3O

2O

I0

/ 1 2 3 4 5 6 7 8 9 I0

Dnscriminetions

Fig. 4. Preoperative (top) and postoperative (bottom) per- formance of rats with piriform cortex lesions (PIRIFORM, n = 4) and piriform sham lesions (SHAM, n = 3) in succes- sive discrimination problems with pairs of different odors. A maximum of 100 trials per day were presented in a session. Group performances were calculated as mean number of trials (mean + S.E.M.) to reach a criterion of 8 correct

choices in 10 consecutive runs.

difficult for animals with a partially damaged piriform cortex,_ we started to use complex odor pairs. Thus, with the exception Of problem 7, all remaining discriminations consisted of two very similar odors that differed in only one chemical (for details see section 'Odors' above).

As Fig. 4 (bottom) illustrates, rats with piriform cortex damage were consistently slower in solving complex problems than control animals, but they were always able to meet the learning criterion; however they learned faster when simple odors had to be discriminated. This finding stands in marked contrast to Expt. 3 where untrained rats with piriform cortex lesions had consistently very high scores or showed no sign of learning in the majority of problems. Post-hoc Scheffe tests revealed significantly impaired performance (p < 0.05) of rats with piriform cortex damage in all but problems 3, 4 and 7.

Histology Analysis of the sections revealed that the D M N

lesion destroyed lateral and medial aspects of the dorsomedial nucleus, the intermediodorsal nucleus as well as parts of the interanteromedial, centromedial, paraventricular and laterodorsal

thalamic nuclei, the stria medullaris, medial and lateral habenular nuclei as well as minor parts of the dorsal hippocampus.

In half of the thalamic control animals the D M N was found to be destroyed, but never com- pletely (mostly the lateral aspect); the inter- mediodorsal nucleus was, however, intact. Fur- ther damaged were paraventricular, paratenial, paracentral, centromedial, interanteromedial, laterodorsal, anterodorsal, anteroventral and anteromedial thalamic nuclei and parts of the posterior thalamic group, the stria medullaris as well as medial and lateral habenular nuclei, the subfornical organ, posterior and habenular com- missure and parts of central gray, anterior pretectal area and dorsal hippocampus.

Analysis of the piriform cortex lesions revealed damage of the LOT starting at the level of the posterior anterior olfactory nucleus and extending caudally to the anterior commissure. In addition, parts of the primary olfactory cortex and olfactory tubercle were destroyed. Some rats had additional damage more rostral in the dorsal, ventral, lateral and medial posterior olfactory nuclei. Their per- formance was indiscriminable from the remaining rats with piriform cortex ablation.

DISCUSSION

Effect of DMN lesions on learning Animals with bilateral damage of the D M N

exhibited a severe deficit in solving olfactory dis- criminations. In contrast, control animals with extensive thalamic lesions that spared the D M N performed virtually indistinguishable from sham lesioned animals, except for an initial learning impairment which was most likely due to un- specific effects of the lesion. Histological analysis revealed that the two types of thalamic lesions overlapped, especially in the area surrounding the DMN. In 7 out of 9 thalamic controls the D M N was destroyed along the exterior border. Appar- ently the lateral aspect of the D M N was not critical in producing a learning impairment. The central area between the ipsi- and contralateral DMN, the intermediodorsal nucleus, was left intact in thalamic control animals. In contrast, this nucleus was destroyed in all but two D M N

124

/~-r----A Y----T--- -.’ \

Fig. 5. Reconstruction of coronal sections (plates 16-25 from ref. 13) illustrating an averaged-size ablation of the DMN.

125

t 1 t

Fig. 6. Reconstruction of coronal sections (plates 16-25 from ref. 13) of an averaged-size lesion of thalamic nuclei adjacent to the D M N (thalamic control).

126

't

Fig. 7. Reconstruction of coronal sections (plates 7-16 from ref. 13) showing an averaged-size ablation of the piriform

cortex.

animals. In these two animals the DMN was completely destroyed unilaterally but only slightly on the opposite side. In addition, part of the inter- mediodorsal nucleus was left intact. Interestingly, performance of these rats was slightly superior compared to the remaining DMN animals. This indicates that the degree of impairment seen after DMN ablation is correlated with the extent of bilateral damage of the medial part of the DMN including the intermediodorsal nucleus. Whether damage of both nuclei together or destruction of the intermediodorsal nucleus alone was responsi- ble for the observed learning impairment cannot be concluded from our data. Interestingly, data exist that show a correlation between a deficit in complex olfactory learning and ablation of the central part of the DMN ~s.

Both preoperatively trained and naive DMN animals demonstrated very poor acquistion of an odor discrimination set with 8 problems that were each presented for 25 trials. A minority of animals was able to reach the learning criterion in individual problems. Several possibilities may account for this inability to learn. (1)The DMN could be the storage site for olfactory cues. How- ever, this assumption can almost certainly be excluded since the DMN is a non-cortical tissue with too small a number of neurons to store innumerable odors. (2)The DMN could be the storage site for procedural memory and would therefore be required to learn a procedure as well as to perform a previously acquired rule. There is evidence indicating a role of the DMN in rule learning TM. (3) The DMN could play a role in the neural mediation of emotional factors such as arousal, attention or motivation, a possibility that has been suggested in earlier studies on the behavioral effects of medial thalamic lesions (e.g. ref. 25). (4)The DMN could be a link in the neural chain of olfactory information processing. Its ablation would therefore cause a direct and specific learning impairment.

Analysis of the results obtained in the third experiment shed some light on the causes of the learning deficit observed in the previous experi- ments. In the first few days of this experiment no sign of learning was evident in DMN animals even though training sessions were significantly

127

longer (100 trials per day). However, all animals successfully learned the third discrimination, and once the learning criterion had been reached less training was necessary to acquire the following problems. Apparently the absolute number of trials given within a day had been critical for the initial learning. Later, a score of 20-30 trials was typical for DMN animals as opposed to a score of 2-5 trials for control animals. Although DMN animals were consistently somewhat slower than sham-lesioned animals their scores did not differ significantly in the last 3 discriminations (post-hoc Scheffe test; this result could be due to the small number of animals per group).

Thus, the behavioral effects seen after DMN lesions reflect a procedural deficit which, how- ever, is temporary, provided intensive training is offered (in contrast to Expt. 3, the standard training schedule used in the first two experiments was insufficient to overcome this procedural deficit). It appears that elevated activity levels and attentional disturbances led to this inability to learn and perform the required procedural steps: analysis of choice sequences within individual sessions of Expt. 3 suggested that the unspecific learning impairment was reflected as 'pattern running' behavior, i.e. choosing arms in a predictable sequence (e.g. running clockwise, alternating between arms). At the beginning of postoperative training 'pattern running' was fre- quently seen in all animals. Control animals typi- cally abandoned this rigid strategy in the course of the first discrimination and started to perform correctly, whereas DMN animals tended to con- tinue during the next 2-4 problems. Later, 'pattern running' was rarely observed and during limited parts of the session. Nevertheless, DMN animals frequently committed only intermittent strings of correct choices during extended periods of random performance, an observation suggest- ing that now a non-procedural impairment was causing retarded learning. In general, the severity of this learning deficit was light and somewhat variable, probably due to the fact that certain odors were harder to learn than others.

In addition, tran sfer of savings occurred incon- sistently when the same odor pair was repeatedly presented within a few days or on consecutive

days (e.g. mean group scores in problems 3, 4 and 5 increased from day 1 to day 2; the amount of errors committed in problem 7 was not reduced compared to problem 4 even though the two problems were identical (see Fig. 3).

In brief, the effect of DMN lesions on olfactory discrimination learning could be summarized as a strong but temporary procedural impairment as well as a lasting encoding deficit which tends to cause anterograde amnesia.

Alternatively, the variety of effects seen after DMN ablation could simply reflect a disturbance in the regulation of arousal and/or motivation (e.g. ref. 25), a deregulation that is most dramatic in an early postlesion phase but then progressively diminishes. This interpretation of the data seems less likely, since animals in Expts. 1 and 2 showed no consistent sign of learning after 9 days of postoperative training.

It is further possible that an increased sensory threshold is responsible for the observed retarded learning. If this was the case one would, however, expect an approximately equal degree of deficit in all problems. Furthermore, it has been shown that damage of the DMN has no impact on odor detection 1.

One study ~8 suggests that damage of the DMN causes an impairment in learning the rule involved in solving a series of olfactory problems. In this view the encoding deficit observed in Expt. 3 reflects an inability to learn a cognitive skill, i.e. the 'learning set' rule. However, the fact that generally no retention of specific odors was observed 24 h after reaching the learning criterion argues for an encoding impairment. Thus, the inability to acquire learning set rules is possibly a consequence of a deficit in encoding olfactory cues.

Memory disturbances in human diencephalic dam- age vs rat DMN lesion

Humans with diencephalic damage are typi- cally impaired in acquiring new and specific information 2°. This anterograde amnesia appears to be the result of a deficit in the initial encoding of information. Similarly, we found that animals with DMN lesions acquire the majority of the discriminations slower than control animals. In

128

addition, periods of correct choices were typically interrupted by random performance. This kind of deficit points to a failure occurring very early in the information processing chain.

In humans, the anterograde amnesia is incom- plete, since the patients are still able, with practice, to acquire problems that require application of rules or strategies to be solved efficiently 2°. For example, the reading of words reversed by a mirror is mastered by these patients, except that the words read are not remembered. Similarly, D M N rats correctly perform all the steps required to run the olfactory maze, even though they do not seem to reliably remember specific odors and their significance learned a day or several days before. Normal rats are able to recall odors for weeks 2~.

To summarize, it appears that the widely held notion that the D M N plays an important role in the establishment of memory of humans at the time of learning also applies for olfactory memory in rats. Thus, further behavioral studies that employ the olfactory modality might give us more insight into the specific role of this nucleus in memory formation in humans.

Effect o f piriform cortex lesions on learning Lesioning the primary projection area of the

olfactory system, the piriform cortex, severely reduced the ability of rats to acquire olfactory discriminations. Apparently, efficient learning of the procedure involved to run the olfactory maze requires an intact piriform cortex. The severity of the deficit was correlated with the extent of bilateral damage, specifically of the LOT. Two rats with a bilateral lesion that was very small on one side and only bordered the L O T met the learning criterion more often than rats equally damaged on both sides.

Training before the lesion dramatically reduced the deficit to a performance almost comparable to control animals. Thus, after preoperative training, a partially intact piriform cortex was sufficient for learning. However , when complex as opposed to simple odors had to be discriminated, rats with piriform cortex damage still made substantially more errors than controls. This points to a failure in efficiently recognizing and remembering com- plex olfactory cues. It has been proposed that

complex olfactory cues require quantitatively more functional connections than simple smells for a neural representation to be formed 7. In this view the observation that acquisition of complex as opposed to simple odors created substantial difficulties is perhaps not surprising.

ACKNOWLEDGEMENT

This research was supported by Gran t (O N R N00014-84-K-0391) from the Office of Naval Research.

REFERENCES

1 Eichenbaum, H., Shedlack, K.J. and Eckmann, K.W., Thalamocortical mechanisms in odor-guided behavior. 1. Effects of lesions of the mediodorsal thalamic nucleus and frontal cortex on olfactory discrimination in the rat, Brain, Behav. Evol., 7 (1983) 255-275.

2 Gross, C.G., Chorover, S.L. and Cohen, S.M., Caudate, cortical, hippocampal and dorsal thalamic lesions in rats. Alternation and Hebb-Williams maze performance, Neuropsyehol., 3 (1965) 53-68.

3 Haberly, L.B. and Price, J.L., Association and com- missural fiber systems of the olfactory cortex of the rat. I. Systems originating in the piriform cortex and adjacent areas, J. Comp. Neurol., 178 (1978) 711-740.

4 Harlow, H.F., The formation of learning sets, Psychol. Rev., 56 (1949) 51-65.

5 Hjorth-Simonsen, A. and Jeune, B., Origin and termina- tion of the hippocampal perforant path in the rat studied by silver impregnation, J. Comp. Neurol., 144 (1972) 215-232.

6 Krettek, J.E. and Price, J.L., The cortical projections of the medio-dorsal nucleus and adjacent thalamic nuclei in the rat, J. Comp. Neurol., 171 (1977) 157-192.

7 Lynch, G., Synapses, Circuits, and the Beginnings of Memory, MIT, Cambridge, 1986.

8 Milner, B., Disorders of learning and memory after temporal lobe lesions in man, Clin. Neurosurg., 19 (1972) 421-446.

9 Nigrosh, B.J., Slotnick, B.M. and Nevin, J.A., Olfactory discrimination, reversal learning, and stimulus control in rats, J. Comp. Physiol. Psychol., 89 (1975) 285-294.

10 O'Keefe, J. and Nadel, L., The Hippocampus as a Cognitive Map, Oxford Univ., London, 1978.

11 Olton, D.S., Becker, J.T. and Handelmann, G.E., Hip- pocampus, space and memory, Behav. Brain Sci., 2 (1979) 313-365.

12 Olton, D.S. and Isaacson, R.L., Effects of lateral and dorsomedial thalamic lesions on retention of active avoidance tasks, J. Comp. Physiol. Psychol., 64 (1967) 256-261.

129

13 Paxinos, G. and Watson, C., The Rat Brain in Stereotaxic Coordinates. Academic, New York, 1982.

14 Price, J.L. and Slotnick, B.M., Dual olfactory representa- tion in the rat thalamus: an anatomical and electro- physiological study, J. Comp. Neurol., 215 (1983) 63-77.

15 Roman, F., Staubli, U. and Lynch, G., Evidence for synaptic potentiation in a cortical network during learn- ing, Brain Res., in press.

16 Slotnick, B.M., Olfactory discrimination in rats with anterior amygdala lesion, Behav. Neurosci., 99 (1985) 956-963.

17 Slotnick, B.M. and Berman, E.J., Transection of the lateral olfactory tract does not produce anosmia, Brain Res. Bull., 5 (1980) 141-145.

18 Slotnick, B.M. and Kaneko, N., Role of mediodorsal thalamic nucleus in olfactory discrimination learning in rats, Science, 214 (1981) 91-92.

19 Slotnick, B.M. and Katz, H.M., Olfactory learning-set formation in rats, Science, 185 (1974) 796-798.

20 Squire, L., The neuropsychology of human memory, Annu. Rev. Neurosci., 5 (1982) 241-273.

21 Staubli, U., Fraser, D., Faraday, R. and Lynch, G., Olfaction and the 'data' memory system in rats, Behav. Neurosci., in press.

22 Staubli, U., Fraser, D., Kessler, M. and Lynch, G., Studies on retrograde and anterograde amnesia of olfactory memory after denervation of the hippocampus, Behav. Neural Biol., 46 (1986) 432-444.

23 Staubli, U., Ivy, G. and Lynch, G., Hippocampal denerva- tion causes rapid forgetting of olfactory information in rats, Proc. Natl. Acad. Sci. U.S.A., 81 (1984) 5885-5887.

24 Staubli, U., Roman, F. and Lynch, G., Selective changes in a cortical network by behaviorally relevant electrical stimulation, Soc. Neurosci. Abstr., 11 (1985) 837.

25 Waring, A.E. and Means, L.W., The effect of medial thalamic lesions on emotionality, activity, and discrimina- tion learning in the rat, Physiol. Behav., 17 (1976) 181-186.