- Email: [email protected]
The choose-short effect in rat memory for event duration: the subjective-shortening model
Behavioural Processes 56 (2001) 31 – 40 www.elsevier.com/locate/behavproc
The choose-short effect in rat memory for event duration: the subjective-shortening model Pascale Leblanc, Monique Soffie´ * Psychobiology Unit-Uni6ersity of Lou6ain 10, Place du Cardinal Mercier, 1348 Lou6ain-la-neu6e, Belgium Received 22 December 1999; received in revised form 8 June 2001; accepted 11 June 2001
Abstract Two predictions derived from the subjective-shortening model were tested in rats. The predictions concerned the temporary occurrence of the choose-short effect with extended training at a given retention interval (RI) and the occurrence of a temporary choose-long effect, when RIs shorter than those used during training were applied. In a first experiment, using a stepwise delay procedure with training 0-s RI sessions interpolated between each series of increasing RIs, results showed: (1) a choose-short effect during the stepwise increase in the delay procedure, (2) a temporary occurrence of the choose-short effect during testing at a given RI and (3) a choose-long effect in half of the animals, when a RI shorter than that used previously was applied. These contrasting results suggest that the disappearance of the choose-short effect could be, as proposed by the model, either the consequence of the foreshortening of the reference memory (for rats choosing-long) or the consequence of an adaptation of the working memory (for rats which did not choose long). Results were discussed in relation with the procedure which could have contributed, by the interposition of 0-s RI sessions, to maintain a stable reference memory. In order to test this interpretation, a second experiment, using the classical stepwise delay procedure without training sessions interpolated, was carried out. In these conditions, rats did never present a significant choose-long effect when the RI was shortened. These results suggest that rats maintained a stable reference memory and could improve their performances during retention testing sessions either by an adaptation of their working memory or by the adoption of an alternative strategy which consisted in learning to maintain an orientation towards the location of the correct lever. © 2001 Elsevier Science B.V. All rights reserved. Keywords: Delayed symbolic matching-to-sample; Working memory; Reference memory; Timing; Rodent
1. Introduction Several studies have used a delayed symbolic matching-to-sample task (DSMTS) to investigate * Corresponding author. Tel.: + 32-10-474091; fax: +3210-473774. E-mail address: [email protected] (M. Soffie´).
memory for event duration. DSMTS involves the presentation of a sample stimulus (e.g. a short or long event duration) followed by a retention interval (RI) and by the presentation of two or more comparison stimuli, one which physically matches the sample. Responses to the comparison stimulus that matches the sample stimulus are rewarded.
0376-6357/01/$ - see front matter © 2001 Elsevier Science B.V. All rights reserved. PII: S 0 3 7 6 - 6 3 5 7 ( 0 1 ) 0 0 1 8 1 - 4
32
P. Leblanc, M. Soffie´ / Beha6ioural Processes 56 (2001) 31–40
When a RI was applied, a pronounced asymmetry in the retention functions has been observed between short- and long-sample trials. Specially, when the RI was lengthened, accuracy decreased more markedly on long-sample trials than on short-sample trials. This effect called the chooseshort effect has been observed on numerous occasions in young adult pigeons (Spetch and Wilkie, 1982, 1983; Spetch and Treit, 1984; Kraemer et al., 1985; Spetch, 1987; Fetterman and MacEwen, 1989; Spetch and Rusak, 1989; Spetch and Sinha, 1989; Grant and Spetch, 1991; Santi et al., 1992; Spetch and Rusak, 1992a,b; Grant, 1993; Grant and Spetch, 1993; Santi et al., 1993; Grant and Spetch, 1994; Fetterman, 1995; Roberts et al., 1995; Grant and Kelly, 1996; Grant et al., 1997; Grant and Kelly, 1998), in only few works in rats (Church, 1980; Al Zahrani et al., 1996; Leblanc and Soffie´ , 1999; Soffie´ et al., 1999) and also in humans (Wearden and Ferrara, 1993). Two models have been proposed to account for the choose-short effect: According to the coding model (Kraemer et al., 1985), the sample durations would be coded into prospective representations on a non-time dimension and the choose-short effect would represent a guessing bias that occurs whenever the code is forgotten. According to the subjective-shortening model proposed by Spetch and Wilkie (1983), the choose-short effect is accounted for by assuming that sample duration is coded analogically and that the choose-short effect arises when subjects retrospectively remember a sample duration that has become foreshortened over the RI. Thus, after a long sample and a long RI, the working memory representation of the long sample would correspond more closely to the reference memory representation of a short sample and would produce an increased tendency to choose ‘short’. This phenomenon is based on the idea that the processes of working memory interact with the processes of reference memory (Honig, 1978). The reference memory of the sample durations and their associations with the comparison stimuli are established during initial training. This reference memory would remain relatively stable when trials with 0-s RI interspersed trials with longer RI (the variable-delay procedure), whereas the work-
ing memory of the sample durations undergoes a systematic change during trials with a RI different from 0-s. On the opposite, the authors suggested that the reference memory should change when the RI was manipulated in a stepwise fashion and should not remain stable on all trials for each RI. In that case, a new reference memory should develop gradually at each RI (i.e. a reference memory which should be foreshortened with extended training at a fixed RI). As a result, the discrepancy between the working memory and the reference memory should diminish and the choose-short effect should disappear. Consequently, when a RI shorter than that employed in extended training was subsequently applied, a choose-long effect should occur because the sample duration represented in working memory should appear temporary subjectively longer than the duration stored in the new foreshortened reference memory. In a previous experiment, using a procedure which differed from the classical stepwise delay procedure by the fact that 0-s RI sessions were interspersed between each series of RI sessions (Leblanc and Soffie´ , 1999), the first prediction derived from the subjective-shortening model was confirmed. Indeed, a choose-short effect and a shift of the psychophysical function relating probability of choosing long to sample durations toward longer durations were shown with increasing RI, in young rats. In the present paper, we attempted to precise the outcome of short-term memory for event durations (i.e. the subjectiveshortening) in young rats by testing the other predictions of the model. The predictions concerned the temporary occurrence of the chooseshort effect (its disappearance) with extended training at a given RI and the occurrence of a choose-long effect, when RIs shorter than those used during training were applied.
2. Experiment I In a first experiment, we tested the two predictions using a procedure close to the procedure adopted in our previous experiment.
P. Leblanc, M. Soffie´ / Beha6ioural Processes 56 (2001) 31–40
2.1. Method 2.1.1. Subjects The subjects were young (6-month old) naı¨ve male Wistar (LOU/C) rats bred in our laboratory. They were grouped two or three per cage in the animal room (temperature: 21– 23 °C) and maintained on a 12/12 normal light– dark schedule. Fifteen days before the first shaping sessions, the rats were deprived of food until they had reached 85% of their free-feeding body weight. They were maintained at that weight throughout the entire experiment. The initial average body weight in grams 9S.E.M. was 31893.04 before the deprivation of food. Fifteen rats were trained until 5-s RI. Six rats presented a persistent choose-short effect beyond twenty sessions at 5-s RI and were discarded from the results. 2.1.2. Apparatus Nine identical skinner boxes (32× 32 ×32 cm3) were used. Each box was placed in an isolation enclosure equipped with a small observation window and each contained two levers ((A) left position and (B) right position) separated by 12 cm, projecting through the front panel, 3 cm above the stainless steel bar floor. A dipper delivered : 0.1 ml of sweetened condensed milk (the reinforcement) through an opening located between the two levers, in the middle of the front panel. Three lights placed on the roof panel and another houselight on the roof of the isolation enclosure were the sources of illumination. The auditory signal was provided by a mechanism located above the dipper. An IBM-PC computer controlled the experimental programme and recorded the data. 2.1.3. Procedure The procedure was the delayed symbolicmatching-to-sample task (DSMTS). A bimodal signal (light and sound) was used; the light was a reduction of luminosity from 50 to 15 lux and the auditory signal was a sound of 50– 60 dB. The procedure was similar to that used in our previous experiment with two exceptions imposed by the predictions which will be tested in the present experiment: (1) the animals were overtrained at
33
the higher 5-s RI and (2) they were then submitted to a stepwise decrease in the delay (i.e. at a lower RI up to 0-s).
2.1.3.1. Training. After the rats had been shaped to respond on both levers (Leblanc et al., 1996), they were submitted at each trial to a signal duration of either 2 or 10 s (sample stimulus) with an equal probability. The end of the signal was followed immediately by the insertion of the levers (comparison stimulus). The rats were trained to press one lever if the signal lasted 2 s and the other if it lasted 10 s. The association of the length of the sample with a particular lever (short sample duration with Lever A and long sample duration with Lever B or vice-versa) was organized so that 50% of the animals had one arrangement and 50% the other. After each correct response, a reinforcement was delivered for 4 s. If an incorrect choice was made, the levers were immediately withdrawn without reinforcement. Another trial began after a 30-s intertrial interval (ITI). A 60-min session was carried out five days a week. Each subject was trained under this procedure until matching accuracy appeared stable and asymptotic. The criterion was reached when the probability of a correct response was significant (PB 0.05, the binomial test) for both levers with a less than 10% difference in performance between levers A and B, during three consecutive sessions. 2.1.3.2. Retention testing. Following training sessions, the rats were submitted successively to phase 1 and to phase 2. Phase 1 : Increases in the RI: During this phase, the RIs were successively increased to 1, 2, 3, 4-s RI, for three consecutive sessions, and for a variable number of sessions at 5-s RI. Within each RI series, the length of the interval remained constant on all trials. One or several training sessions (0-s RI) were interposed between each RI series, until the rats recovered the criterion performance level, for one session. Finally, at the last 5-s RI, the rats were trained until they reached the criterion performance level for three sessions. Afterwards, they were submitted to phase 2.
34
P. Leblanc, M. Soffie´ / Beha6ioural Processes 56 (2001) 31–40
Phase 2 : Decrease in the RI: During this phase, the RI was decreased from 5 to 0-s. The 0-s RI was maintained during three consecutive sessions.
2.2. Results As shown in Fig. 1, during the three individual sessions following each increase in the RI from 2 to 5-s, all rats showed a consistent choose-short effect, which was confirmed by significant differences between short and long correct responses [t-test: (RI 2-s): t(8)=2.43, P B 0.05; (RI 3-s): t(8) =3.16, P =0.01; (RI 4-s): t(8) = 6.06, P B 0.001; (RI 5-s): t(8)= 4.93, P B 0.001]. With extended training at a fixed RI, the choose-short effect disappeared. Indeed, no significant difference between the two durations was found when the last block of three sessions at 5-s RI was taken into account. To test the significance of the decrease of the choose-short effect with extended testing at 5-s RI, the first and the last blocks of three sessions at 5-s RI were compared by t-tests. The tests were computed on the mean difference and also on the ‘choice ratios’. The ‘choice ratios’ were calculated
Fig. 1. Mean percentage of correct responses after short- and long-signal durations during the three individual sessions after each change in the RI and during three last sessions at 5-s RI. The last block represents the performances of the two subgroups [rats which presented (N =5) or did not presented (N =4) the choose-long effect]. The horizontal line indicates the significant threshold of correct responses (binomial test).
by dividing the percentage of correct choices after short-signal duration by the sum of the percentage of correct choices after both short- and longsignal durations. Ratios greater than 0.5 indicate higher accuracy after short-signal duration (i.e., the choose-short effect) and ratios of less than 0.5 indicate higher accuracy after long-signal duration (i.e., the choose-long effect). The prediction that a decrease in the choose-short effect would be reflected by a decrease in the mean difference [t(8)= 2.69, P= 0.01)] and in the ‘choice ratio’ was confirmed [t(8)= 2.74, P = 0.01]. The choice ratios decreased from a mean value of 0.59 to a mean value of 0.52. The mean number of sessions (9 S.E.M.) needed to observe the choose-short effect disappearance was 9.559 1.33. As far as the three sessions following the decrease of the RI from 5 to 0-s were concerned, when the nine rats were taken into account, no significant preference (choose-long or chooseshort) appeared and only one of nine rats reached the criterion performance level. In order to test whether the appearance of the choose-short effect, after each increase in the RI and the appearance of the choose-long effect, after a decrease in the RI to 0-s., were higher during the first session after each change in the RI, data analyses were computed only on the first session. When the choose-short effect was analysed, results showed similar but not as strong significant differences between short and long correct responses than those obtained when the three sessions were taken into account. The chooseshort effect appeared in all rats on and after 2-s RI [t-test: (RI 2-s): t(8)= 2.66, P B 0.05; (RI 3-s): t(8)= 3.65, PB 0.01; (RI 4-s): t(8)= 6.84, PB 0.001; (RI 5-s): t(8)= 4.04, P B 0.01]. The chooselong effect was again not significant. However, several significant effects emerged when the animals were split into two subgroups; one group included the five rats which presented more long than short correct responses and the other group included the remaining four animals which presented either an equal proportion of short and long correct responses or even a higher proportion of short correct responses. When statistical analyses were computed on the first group of five rats, a significant choose-long effect
P. Leblanc, M. Soffie´ / Beha6ioural Processes 56 (2001) 31–40
appeared whatever the three sessions [t(4) = 3.87, P B0.05] or only the first session [t(4) = 2.97, P B0.05] were taken into account. This effect never persists beyond five sessions. When the analyses were computed on the other group, no significant results appeared. The separation of experimental rats into two subgroups can not be explained by a procedural artefact such as a positional bias. As reported in the procedure, the position of the lever associated with the long duration was counterbalanced across the nine rats and the assignment of the levers remained random and did not differ into the two subgroups. Although the absence of procedural artefacts, the occurrence of choose-long responses can not be predicted by the performances of the animal during the former sessions. Indeed, never significant correlations were found between the preference for long response and either the ability to learn the duration discrimination (number of sessions to reach the criterion during training) or the extend or the persistence of the choose-short effect at 5-s RI, expressed, respectively by the differences between short or long correct responses and by the number of sessions to observe the disappearance of the choose-short preference.
2.3. Discussion The present experiment confirms our previous study (Leblanc and Soffie´ , 1999) having reported a choose-short effect in rats and confirms one of the above mentioned predictions, derived from the subjective-shortening model (Spetch and Wilkie, 1983). Indeed, the temporary occurrence of the choose-short effect was verified by the significant decrease in the mean difference between short and long correct responses and by the significant decrease in the choice ratios, observed when the first and the last blocks of three sessions at 5-s RI were compared. However, the choose-long effect was found only in five out of nine rats, when the RI was decreased to 0-s., whatever was the lever assignment. This effect disappeared more rapidly than the choose-short effect but persisted on and over the first session. No difference appeared in the
35
strength of the occurrence of both the chooseshort and choose-long effects, when only the first session after each change in the RI was taken into account. According to Spetch and Wilkie (1983), Spetch (1987), Grant et al. (1997), the choose-long effect occurred because a new reference memory, based on the working memory, would be formed at a fixed RI and because the working memory of the samples did not have been sufficiently foreshortened when the RI was decreased from 5 to 0-s. The absence of a choose-long effect in the remaining four rats, suggests that the reference memory would remain stable. It would be the content of the working memory which would become progressively close to the content of the reference memory and which would explain the disappearance of the choose-short effect. This supposed that the content of the working memory did not remain foreshortened, when a fixed RI was applied. The procedure used could be proposed to account for these contrasting results. Indeed, in our procedure, contrary to the classical stepwise delay procedure, 0-s training sessions were interposed between each series of the successive retention sessions in phase 1 (increases in the RI). It could be hypothesized that this modification of the procedure should have contributed to maintain a stable reference memory. In order to test this interpretation, a second experiment was carried out, using the classical stepwise delay procedure.
3. Experiment II In this experiment, young rats were trained in a classical stepwise delay procedure, without receiving 0-s training sessions between each RI series and with overtraining at each RI (Spetch and Wilkie, 1983).
3.1. Method 3.1.1. Subjects The subjects were young (6-month old) naı¨ve male Wistar (LOU/C) rats bred, reared and food deprived similarly as described in experiment I. At
36
P. Leblanc, M. Soffie´ / Beha6ioural Processes 56 (2001) 31–40
the beginning of the experiment, there were fourteen rats. As in experiment I, six rats which present a persistent choose- short effect beyond 20 sessions either at 4-s RI (N =4) or at 5-s RI (N = 2) were discarded and the data of only eight animals were analysed. The initial average body weight in grams9S.E.M. was 388.919 8.6 before the deprivation of food.
3.1.2. Apparatus The apparatus was the same as that used in Experiment I. 3.1.3. Procedure The procedure was similar to that used in Experiment I with two exceptions: (1) 0- s training sessions were not interposed between each RI series and (2) the animals were overtrained at each RI until the choose-short effect was disappeared. 3.1.3.1. Training. The training sessions were the same as that used in Experiment I 3.1.3.2. Retention testing. Following training sessions, the rats were submitted successively to phase 1 and to phase 2. Phase 1: Increases in the RI: During this phase, the RI was successively increased, for several consecutive sessions, to 1, 2, 3, 4 and 5-s RI. Within each RI series, the length of the interval remained constant on all trials. The rats were trained until they reached the criterion performance level for three sessions. Afterwards, they were submitted to phase 2. Phase 2: Decrease in the RI: During this phase, the RI was decreased from 5 to 0-s and the 0-s RI was maintained for three consecutive sessions, as in experiment I.
4. Results As shown in Fig. 2, during the first three individual sessions following each increase in the RI from 1 to 5-s, the choose-short effect was less consistent than in experiment I. It was only significant at three RIs and was not proportional to the retention delay. [t-test: (RI 1-s): t(7) = 2.73,
Fig. 2. Mean percentage of correct responses after short- and long-signal durations during the three first and the three last individual sessions after each change in the RI. The horizontal line indicates the significant threshold of correct responses (binomial test).
PB 0.05; (RI 3-s): t(7)=2.62, PB 0.05; (RI 4-s): t(7)= 2.9, PB0.05]. Moreover, when only the first session at each RI was taken into account, no choose-short effect appeared. With extended training at each RI, the chooseshort effect disappeared. To test the significance of the decrease of the choose-short effect at each RI, the first and the last blocks of three sessions were compared by t-tests. The tests were computed on the mean difference and also on the ‘choice ratios’. The decrease in the choose-short effect was expressed by a decrease in the mean difference [t-test: (RI 1-s): t(7)= − 4.48, P= 0.01; (RI 3-s): t(7)= − 3.57, P=0.01(RI 4-s): t(7)= − 3.17, P B0.05] and in the ‘choice ratio’ [t-test: (RI 1-s): t(7)= − 4.09, P= 0.01; (RI 3-s): t(7)= − 2.93, PB 0.05; (RI 4-s): t(7)= − 2.82, PB 0.05]. The mean number of sessions needed to observe a significant decrease at the different RIs were: 6.6 (RI 1-s), 4 (RI 2-s), 2.2 (RI 3-s), 6 (RI 4-s) and 9 (RI 5-s). As far as the three sessions following the decrease of the RI from 5 to 0-s were concerned, when the eight rats were taken into account, no significant choose-long effect appeared. The matching accuracy for both durations (i.e. the criterion level) reached during the last 5-s RI sessions was maintained in seven out of eight rats.
P. Leblanc, M. Soffie´ / Beha6ioural Processes 56 (2001) 31–40
Finally, a comparison of the percentage of correct responses between the two experiments did not show significant differences at each RI whatever the first or three first sessions were taken into account.
37
at the time of the duration sample and consequently, during extended training at a given RI, the rat could progressively learn to adopt a behavioural successful strategy which consisted in learning to maintain an orientation to the location of the correct lever over the RI.
5. Discussion 6. General discussion When training sessions were not interspersed between each RI series, as in the first experiment, the choose-short effect was less consistent. It appeared only when the three sessions at each RI were taken into account. These results are consistent with the procedure. It is obvious that an increase of 1-s between each RI was less disturbing than the increases of the RI up to 5-s imposed in experiment I. With extended training at each RI, the choose-short effect disappeared in the majority of rats, in which performances reached the criterion level. These results convey that the discrepancy between the working memory and the reference memory should have disappeared. Moreover, the hypothesis concerning the incidence of the procedure, proposed in the discussion of experiment I, was not verified. The absence of a significant choose-long effect and the recovery of accurate performances when the RI was decreased from 5 to 0-s suggested that the content of the reference memory would remain stable even when no training sessions were interspersed between each RI. Given that there was no evidence that the reference memory underwent a systematic change within the RI trials, the recovery of accurate performances with extended training during retention testing sessions should not be ascribed, as in pigeon’s studies, to a foreshortening of the reference memory. As proposed in experiment I, the disappearance of the chooseshort effect coming with the recovery of accurate performances could be, the consequence of an adaptation of the working memory which did not remain shortened. This interpretation is attractive, but unfortunately another non-memorial interpretation could explain the differences between rats and pigeons adaptations. In our experiment, the location of the correct response can be determined
The two experiments confirm, as already reported (Leblanc and Soffie´ , 1999; Soffie´ et al., 1999), that rats present a choose-short effect when a RI was applied between the offset of sample stimulus and the opportunity to respond. This effect was not often reported in rat studies which, unlike pigeon studies, often reported a chooselong effect (Roberts, 1982; Meck et al., 1984; Berz et al., 1992) and rarely a choose-short effect. Research with rats has shown that the apparition of the choose-short effect was largely influenced by procedural conditions. According to Santi et al. (1995, 1997), the rats evaluated the long duration of the sample as being ‘shorter’ compared to the objective duration in the context of a long ITI and with a change in lighting condition between the presentation of the sample and the ITI. Our experimental conditions (bimodal light/sound stimuli and 30 s-ITI), which were close to Santi’s conditions, could have contributed to promote the choose-short effect in our rats. Previous research, having reported a chooseshort-effect in rats, did not necessarily show a subjective-shortening of event duration. A forgetting which did not occur on time dimension was proposed by Church (1980) to explain the development of a choose-short bias. This interpretation appears unlikely to account for the development of the choose-short effect in our experiments, given that, contrary to Church’s experiment, a shift of the psychological function to the right and an increase of the point of subjective equality (PSE) was shown in our rats in similar experimental conditions (Leblanc and Soffie´ , 1999). According to the confusion/instructional failure hypothesis, proposed by Sherburne et al., (1998), if the ITI and the RI conditions match during test trials, the animal may not readily discriminate
38
P. Leblanc, M. Soffie´ / Beha6ioural Processes 56 (2001) 31–40
between them. Clearly spoken, if the animal had difficulty discriminating between the two intervals, it was uncertain what to do and when the two levers appeared after a RI, the experimental situation could be interpreted like a ‘no sample’ trial. Given that ‘no sample’ is closer to the short than to the long sample stimulus, the animal responded short. The observation of the behavioural strategy of our animals during the retention sessions did not support this confusion hypothesis. Indeed, systematic observations did disentangle two distinct strategies during RI and ITI: during RIs the rat remained in front of the box facing the levers and the feeding trough area, whereas during ITIs it walked around the whole skinner box. Finally, a procedural artefact such as positional response bias (Santi et al., 1995) could neither account for our results given that the chooseshort effect appeared in all rats and that the lever assignment was counterbalanced across our experimental rats. All these elements suggest that the choose-short effect, which appeared in rats in our experimental conditions, was not due to a procedural artefact and do not rule out a memorial factor such as a subjective-shortening. As far as the other predictions derived from the model are concerned, our results are not very conclusive. The predictions attempted to precise the interactive processes between working memory and reference memory involved when the choose-short effect disappeared. The temporary occurrence was verified in most but not in all rats. Given that the disappearance of the choose-short effect corresponded with the recovery of equivalent and significant correct performances for both short and long responses, the content of working memory would be similar to the content of reference memory. Pigeon’s experiments suggested that the disappearance of the choose-short effect was the consequence of the foreshortening of the duration stored in reference memory which became similar to the content of the duration stored in working memory during the current trial with RI. Accordingly, when a shorter RI was subsequently applied, the current sample duration appeared longer than the foreshortened reference memory
and the animal responded long. However, our data show that the majority of rats, instead of exhibiting a choose-long preference, maintained a good performance level when a shorter RI was applied. These results obtained in rats are consistent either with a different memory adaptation or with the development of an alternative behavioural strategy. The memory interpretation would suggest that the pivotal aspect between working memory and reference memory in rats was not the same as in pigeons and could be accounted for as follows: if rats maintained a stable reference memory during the RI sessions, a possibility to reduce the discrepancy between the two memories would be to adapt their working memory. In that case the rats, which had learned to evaluate the short- and the long-sample at the exact value even after a long 5-s RI, would not be disturbed during the subsequent shorter RI trials. During these shorter RI trials (0-s RI in our experiment) the working memory would not yet be involved and the information from timing processes would be directly compared to the durations stored in a stable reference memory. It is well known that rats are very spatial animals whereas pigeons are very visual animals. Accordingly, pigeons are more able than rats to learn a task, in which the correct choice stimulus did not remain in the same location on all trials and, which prevented the determination of the correct choice prior the RI. On the opposite, in rats studies, such as in our experiments, the correct lever associated with each duration remained in the same location and the rat, having the opportunity to know the correct choice prior the RI, can learn to maintain an orientation towards the location of the correct lever over the RI. Consequently, a non-memorial interpretation cannot be ruled out. However, the recuperation of accurate performances for both duration after the longest 4 and 5-s RI seemed not to be easy for the rat. Indeed, when the total of our experimental subjects were taken into account, 12 out of 29 rats did not succeed, even after 20 training sessions, to loose their choose-short preference either at 4 or at 5-s RI and in both experiments. Moreover, for rats which complete the experiments, a mean of, respectively 9.55 and 9 sessions were needed to
P. Leblanc, M. Soffie´ / Beha6ioural Processes 56 (2001) 31–40
recover the criterion performance level at 5-s RI. If rats were able to respond accurately, without the use of an effortful memory process (i.e., a working memory adaptation), but with the use of an alternative orientation strategy, how can we explain that 12 out of 29 rats did not succeed and that the remaining rats which complete the task did not recover the criterion performance level at 5-s RI more quickly? To summarize, the results of the studies expressed in this paper suggest that rats, as well as pigeons, are able to show a forgetting occurring on a time dimension and support the idea that the reference memory remain stable when the chooseshort effect occurred. However, there was no evidence, as in pigeons studies, that in rats the reference memory underwent systematic changes during retention testing sessions. Our results do not succeed to disentangle whether the recuperation of accurate performances during retention trials was the consequence of an adaptation of working memory or the consequence of the development of an alternative spatial strategy.
Acknowledgements We would like to thank Marie Bronchart and Marc-Henri Weyers for supervising the development of the programme of the experiment and data analyses and Christiane Van Vaerenbergh for her active collaboration in testing and data collection.
References Al Zahrani, S.S.A., Ho, M.Y., Velazquez Martinez, D.N., Lopez Cabrera, M., Bradshaw, C.M., Szabadi, E., 1996. Effect of destruction of the 5-hydroxytryptaminergic pathways on the acquisition of temporal discrimination and memory for duration in a delayed conditional discrimination task. Psychopharmacology 23, 103 –110. Berz, S., Ba¨ ttig, K., Welzl, H., 1992. The effects of anticholinergic drugs on delayed time discrimination performance in rats. Physiol. Behav. 51, 493 –499. Church, R.M., 1980. Short-term memory for time intervals. Learn. Motiv. 11, 208 –219. Fetterman, J.G., 1995. The psychophysics of remembered duration. Anim. Learn. Behav. 23 (1), 49 –62.
39
Fetterman, J.G., MacEwen, D., 1989. Short-term memory for responses: the choose-small effect. J. Exp. Anal. Behav. 52, 311 – 324. Grant, D.S., 1993. Coding processes in pigeons. In: Zentall, T.R. (Ed.), Animal cognition: a tribute to Donald A. Riley. Erlbaum, Hillsdale, NJ, pp. 193 – 196. Grant, D.S., Kelly, R., 1996. The role of minimum wait time and sample discriminability in the coding of event duration in pigeons. Learn. Motiv. 27, 243 – 259. Grant, D.S., Kelly, R., 1998. The effect of variable-delay training on coding of event duration in pigeons. Learn. Motiv. 29, 49 – 67. Grant, D.S., Spetch, M.L., 1991. Pigeons’ memory for event duration: differences between choice and successive matching tasks. Learn. Motiv. 22, 180 – 190. Grant, D.S., Spetch, M.L., 1993. Analogical and non-analogical coding of samples differing in duration in a choicematching task in pigeons. J. Exp. Psychol.: Anim. Behav. Process 19, 15 – 25. Grant, D.S., Spetch, M.L., 1994. Mediated transfer testing provides evidence for common coding of duration and line samples in many-to-one matching in pigeons. Anim. Learn. Behav. 22, 84 – 89. Grant, D.S., Spetch, M.L., Kelly, R., 1997. Pigeons’ coding of event duration in delayed matching to sample. In: Bradshaw, C.M., Szabadi, E. (Eds.), Time and Behavior: Psychological and Neurobehavioural Analyses. Elsevier, Amsterdam, pp. 217 – 251. Honig, W.K., 1978. Studies of working memory in the pigeon. In: Hulse, S.H., Fowler, H., Honig, W.K. (Eds.), Cognitive Processes in Animal Behavior. Erlbaum, Hillsdale, NJ, pp. 211 – 248. Kraemer, P.J., Mazmanian, D.S., Roberts, W.A., 1985. The choose-short effect in pigeon memory for stimulus duration: subjective-shortening versus coding models. Anim. Learn. Behav. 13, 349 – 354. Leblanc, P., Soffie´ , M., 1999. Effects of age on short-term memory for time in rats. Exp. Aging Res. 25, 267 – 284. Leblanc, P., Weyers, M.H., Soffie´ , M., 1996. Age-related differences for duration discrimination in rats. Physiol. Behav. 60 (2), 555 – 558. Meck, W.H., Church, R.M., Olton, D.S., 1984. Hippocampus, time and memory. Behav. Neurosci. 98, 3 – 22. Roberts, W.A., 1982. Cross-modal use of an internal clock. J. Exp.Psychol.: Anim. Behav. Process 8, 23 – 32. Roberts, W.A., Macuda, T., Brodbeck, D.R., 1995. Memory for number of light flashes in the pigeon. Anim. Learn. Behav. 23 (2), 182 – 188. Santi, A., Bridson, S., Ducharme, M.J., 1993. Memory codes for temporal and non-temporal samples in many to one matching by pigeons. Anim. Learn. Behav. 21, 120. Santi, A., Ducharme, M.J., Bridson, S., 1992. Differential outcome expectancies and memory for temporal and nontemporal stimuli in pigeons. Learn. Motiv. 23, 156 – 169. Santi, A., Standford, L., Coyle, J., 1997. Rats’memory for event duration: differential effects of delaying the discriminative choice cue as opposed to the opportunity 130 to execute the choice response. Behav.Proc. 40, 193 – 199.
40
P. Leblanc, M. Soffie´ / Beha6ioural Processes 56 (2001) 31–40
Santi, A., Weise, L., Kuiper, D., 1995. Memory for event duration in rats. Learn. Motiv. 26, 83 –100. Sherburne, L.M., Zentall, T.R., Kaiser, D.H., 1998. Timing in pigeons: the choose-short effect may result from pigeons’ ‘confusion’ between delay and intertrial intervals. Psychon. Bull. Rev. 5, 516 – 522. Soffie´ , M., Hahn, K., Terao, E., Eclancher, F., 1999. Behavioural and glial changes in old rats following environmental enrichment. Behav. Brain Res. 101, 37 –49. Spetch, M.L., 1987. Systematic errors in pigeons’ memory for event duration: interaction between training and test delay. Anim. Learn. Behav. 15, 1 –5. Spetch, M.L., Rusak, B., 1989. Pigeons’ memory for event duration: intertrial interval and delay effects. Anim. Learn. Behav. 17, 147 – 156. Spetch, M.L., Rusak, B., 1992a. Temporal context effects in pigeons’ memory for event duration. Learn. Motiv 23 (117), 144.
Spetch, M.L., Rusak, B., 1992b. Time present and time past. In: Honig, W.K., Fetterman, J.G. (Eds.), Cognitive Aspects of Stimulus Control. Erlbaum, Hillsdale, NJ, pp. 47 – 67. Spetch, M.L., Sinha, S.S., 1989. Proactive effects in pigeons’ memory for event duration: evidence against the coding model. J. Exp. Psychol.: Anim. Behav. Proc. 15, 347 – 357. Spetch, M.L., Treit, D., 1984. The effect of d-amphetamine on short-term memory for time in pigeons. Pharmacol. Biochem. Behav. 21, 663 – 666. Spetch, M.L., Wilkie, D.M., 1982. A systematic bias in pigeons’ memory for food and light duration. Behav. Anal. Lett. 2, 267 – 274. Spetch, M.L., Wilkie, D.M., 1983. Subjective-shortening: a model of pigeons’memory for event duration. J. Exp. Psychol.: Anim. Behav. Process. 9, 14 – 30. Wearden, J.K., Ferrara, A., 1993. Subjective shortening in humans’ memory for stimulus duration. Q. J. Exp. Psychol. 46 (2), 163 – 186.
The choose-short effect in rat memory for event duration: the subjective-shortening model Pascale Leblanc, Monique Soffie´ * Psychobiology Unit-Uni6ersity of Lou6ain 10, Place du Cardinal Mercier, 1348 Lou6ain-la-neu6e, Belgium Received 22 December 1999; received in revised form 8 June 2001; accepted 11 June 2001
Abstract Two predictions derived from the subjective-shortening model were tested in rats. The predictions concerned the temporary occurrence of the choose-short effect with extended training at a given retention interval (RI) and the occurrence of a temporary choose-long effect, when RIs shorter than those used during training were applied. In a first experiment, using a stepwise delay procedure with training 0-s RI sessions interpolated between each series of increasing RIs, results showed: (1) a choose-short effect during the stepwise increase in the delay procedure, (2) a temporary occurrence of the choose-short effect during testing at a given RI and (3) a choose-long effect in half of the animals, when a RI shorter than that used previously was applied. These contrasting results suggest that the disappearance of the choose-short effect could be, as proposed by the model, either the consequence of the foreshortening of the reference memory (for rats choosing-long) or the consequence of an adaptation of the working memory (for rats which did not choose long). Results were discussed in relation with the procedure which could have contributed, by the interposition of 0-s RI sessions, to maintain a stable reference memory. In order to test this interpretation, a second experiment, using the classical stepwise delay procedure without training sessions interpolated, was carried out. In these conditions, rats did never present a significant choose-long effect when the RI was shortened. These results suggest that rats maintained a stable reference memory and could improve their performances during retention testing sessions either by an adaptation of their working memory or by the adoption of an alternative strategy which consisted in learning to maintain an orientation towards the location of the correct lever. © 2001 Elsevier Science B.V. All rights reserved. Keywords: Delayed symbolic matching-to-sample; Working memory; Reference memory; Timing; Rodent
1. Introduction Several studies have used a delayed symbolic matching-to-sample task (DSMTS) to investigate * Corresponding author. Tel.: + 32-10-474091; fax: +3210-473774. E-mail address: [email protected] (M. Soffie´).
memory for event duration. DSMTS involves the presentation of a sample stimulus (e.g. a short or long event duration) followed by a retention interval (RI) and by the presentation of two or more comparison stimuli, one which physically matches the sample. Responses to the comparison stimulus that matches the sample stimulus are rewarded.
0376-6357/01/$ - see front matter © 2001 Elsevier Science B.V. All rights reserved. PII: S 0 3 7 6 - 6 3 5 7 ( 0 1 ) 0 0 1 8 1 - 4
32
P. Leblanc, M. Soffie´ / Beha6ioural Processes 56 (2001) 31–40
When a RI was applied, a pronounced asymmetry in the retention functions has been observed between short- and long-sample trials. Specially, when the RI was lengthened, accuracy decreased more markedly on long-sample trials than on short-sample trials. This effect called the chooseshort effect has been observed on numerous occasions in young adult pigeons (Spetch and Wilkie, 1982, 1983; Spetch and Treit, 1984; Kraemer et al., 1985; Spetch, 1987; Fetterman and MacEwen, 1989; Spetch and Rusak, 1989; Spetch and Sinha, 1989; Grant and Spetch, 1991; Santi et al., 1992; Spetch and Rusak, 1992a,b; Grant, 1993; Grant and Spetch, 1993; Santi et al., 1993; Grant and Spetch, 1994; Fetterman, 1995; Roberts et al., 1995; Grant and Kelly, 1996; Grant et al., 1997; Grant and Kelly, 1998), in only few works in rats (Church, 1980; Al Zahrani et al., 1996; Leblanc and Soffie´ , 1999; Soffie´ et al., 1999) and also in humans (Wearden and Ferrara, 1993). Two models have been proposed to account for the choose-short effect: According to the coding model (Kraemer et al., 1985), the sample durations would be coded into prospective representations on a non-time dimension and the choose-short effect would represent a guessing bias that occurs whenever the code is forgotten. According to the subjective-shortening model proposed by Spetch and Wilkie (1983), the choose-short effect is accounted for by assuming that sample duration is coded analogically and that the choose-short effect arises when subjects retrospectively remember a sample duration that has become foreshortened over the RI. Thus, after a long sample and a long RI, the working memory representation of the long sample would correspond more closely to the reference memory representation of a short sample and would produce an increased tendency to choose ‘short’. This phenomenon is based on the idea that the processes of working memory interact with the processes of reference memory (Honig, 1978). The reference memory of the sample durations and their associations with the comparison stimuli are established during initial training. This reference memory would remain relatively stable when trials with 0-s RI interspersed trials with longer RI (the variable-delay procedure), whereas the work-
ing memory of the sample durations undergoes a systematic change during trials with a RI different from 0-s. On the opposite, the authors suggested that the reference memory should change when the RI was manipulated in a stepwise fashion and should not remain stable on all trials for each RI. In that case, a new reference memory should develop gradually at each RI (i.e. a reference memory which should be foreshortened with extended training at a fixed RI). As a result, the discrepancy between the working memory and the reference memory should diminish and the choose-short effect should disappear. Consequently, when a RI shorter than that employed in extended training was subsequently applied, a choose-long effect should occur because the sample duration represented in working memory should appear temporary subjectively longer than the duration stored in the new foreshortened reference memory. In a previous experiment, using a procedure which differed from the classical stepwise delay procedure by the fact that 0-s RI sessions were interspersed between each series of RI sessions (Leblanc and Soffie´ , 1999), the first prediction derived from the subjective-shortening model was confirmed. Indeed, a choose-short effect and a shift of the psychophysical function relating probability of choosing long to sample durations toward longer durations were shown with increasing RI, in young rats. In the present paper, we attempted to precise the outcome of short-term memory for event durations (i.e. the subjectiveshortening) in young rats by testing the other predictions of the model. The predictions concerned the temporary occurrence of the chooseshort effect (its disappearance) with extended training at a given RI and the occurrence of a choose-long effect, when RIs shorter than those used during training were applied.
2. Experiment I In a first experiment, we tested the two predictions using a procedure close to the procedure adopted in our previous experiment.
P. Leblanc, M. Soffie´ / Beha6ioural Processes 56 (2001) 31–40
2.1. Method 2.1.1. Subjects The subjects were young (6-month old) naı¨ve male Wistar (LOU/C) rats bred in our laboratory. They were grouped two or three per cage in the animal room (temperature: 21– 23 °C) and maintained on a 12/12 normal light– dark schedule. Fifteen days before the first shaping sessions, the rats were deprived of food until they had reached 85% of their free-feeding body weight. They were maintained at that weight throughout the entire experiment. The initial average body weight in grams 9S.E.M. was 31893.04 before the deprivation of food. Fifteen rats were trained until 5-s RI. Six rats presented a persistent choose-short effect beyond twenty sessions at 5-s RI and were discarded from the results. 2.1.2. Apparatus Nine identical skinner boxes (32× 32 ×32 cm3) were used. Each box was placed in an isolation enclosure equipped with a small observation window and each contained two levers ((A) left position and (B) right position) separated by 12 cm, projecting through the front panel, 3 cm above the stainless steel bar floor. A dipper delivered : 0.1 ml of sweetened condensed milk (the reinforcement) through an opening located between the two levers, in the middle of the front panel. Three lights placed on the roof panel and another houselight on the roof of the isolation enclosure were the sources of illumination. The auditory signal was provided by a mechanism located above the dipper. An IBM-PC computer controlled the experimental programme and recorded the data. 2.1.3. Procedure The procedure was the delayed symbolicmatching-to-sample task (DSMTS). A bimodal signal (light and sound) was used; the light was a reduction of luminosity from 50 to 15 lux and the auditory signal was a sound of 50– 60 dB. The procedure was similar to that used in our previous experiment with two exceptions imposed by the predictions which will be tested in the present experiment: (1) the animals were overtrained at
33
the higher 5-s RI and (2) they were then submitted to a stepwise decrease in the delay (i.e. at a lower RI up to 0-s).
2.1.3.1. Training. After the rats had been shaped to respond on both levers (Leblanc et al., 1996), they were submitted at each trial to a signal duration of either 2 or 10 s (sample stimulus) with an equal probability. The end of the signal was followed immediately by the insertion of the levers (comparison stimulus). The rats were trained to press one lever if the signal lasted 2 s and the other if it lasted 10 s. The association of the length of the sample with a particular lever (short sample duration with Lever A and long sample duration with Lever B or vice-versa) was organized so that 50% of the animals had one arrangement and 50% the other. After each correct response, a reinforcement was delivered for 4 s. If an incorrect choice was made, the levers were immediately withdrawn without reinforcement. Another trial began after a 30-s intertrial interval (ITI). A 60-min session was carried out five days a week. Each subject was trained under this procedure until matching accuracy appeared stable and asymptotic. The criterion was reached when the probability of a correct response was significant (PB 0.05, the binomial test) for both levers with a less than 10% difference in performance between levers A and B, during three consecutive sessions. 2.1.3.2. Retention testing. Following training sessions, the rats were submitted successively to phase 1 and to phase 2. Phase 1 : Increases in the RI: During this phase, the RIs were successively increased to 1, 2, 3, 4-s RI, for three consecutive sessions, and for a variable number of sessions at 5-s RI. Within each RI series, the length of the interval remained constant on all trials. One or several training sessions (0-s RI) were interposed between each RI series, until the rats recovered the criterion performance level, for one session. Finally, at the last 5-s RI, the rats were trained until they reached the criterion performance level for three sessions. Afterwards, they were submitted to phase 2.
34
P. Leblanc, M. Soffie´ / Beha6ioural Processes 56 (2001) 31–40
Phase 2 : Decrease in the RI: During this phase, the RI was decreased from 5 to 0-s. The 0-s RI was maintained during three consecutive sessions.
2.2. Results As shown in Fig. 1, during the three individual sessions following each increase in the RI from 2 to 5-s, all rats showed a consistent choose-short effect, which was confirmed by significant differences between short and long correct responses [t-test: (RI 2-s): t(8)=2.43, P B 0.05; (RI 3-s): t(8) =3.16, P =0.01; (RI 4-s): t(8) = 6.06, P B 0.001; (RI 5-s): t(8)= 4.93, P B 0.001]. With extended training at a fixed RI, the choose-short effect disappeared. Indeed, no significant difference between the two durations was found when the last block of three sessions at 5-s RI was taken into account. To test the significance of the decrease of the choose-short effect with extended testing at 5-s RI, the first and the last blocks of three sessions at 5-s RI were compared by t-tests. The tests were computed on the mean difference and also on the ‘choice ratios’. The ‘choice ratios’ were calculated
Fig. 1. Mean percentage of correct responses after short- and long-signal durations during the three individual sessions after each change in the RI and during three last sessions at 5-s RI. The last block represents the performances of the two subgroups [rats which presented (N =5) or did not presented (N =4) the choose-long effect]. The horizontal line indicates the significant threshold of correct responses (binomial test).
by dividing the percentage of correct choices after short-signal duration by the sum of the percentage of correct choices after both short- and longsignal durations. Ratios greater than 0.5 indicate higher accuracy after short-signal duration (i.e., the choose-short effect) and ratios of less than 0.5 indicate higher accuracy after long-signal duration (i.e., the choose-long effect). The prediction that a decrease in the choose-short effect would be reflected by a decrease in the mean difference [t(8)= 2.69, P= 0.01)] and in the ‘choice ratio’ was confirmed [t(8)= 2.74, P = 0.01]. The choice ratios decreased from a mean value of 0.59 to a mean value of 0.52. The mean number of sessions (9 S.E.M.) needed to observe the choose-short effect disappearance was 9.559 1.33. As far as the three sessions following the decrease of the RI from 5 to 0-s were concerned, when the nine rats were taken into account, no significant preference (choose-long or chooseshort) appeared and only one of nine rats reached the criterion performance level. In order to test whether the appearance of the choose-short effect, after each increase in the RI and the appearance of the choose-long effect, after a decrease in the RI to 0-s., were higher during the first session after each change in the RI, data analyses were computed only on the first session. When the choose-short effect was analysed, results showed similar but not as strong significant differences between short and long correct responses than those obtained when the three sessions were taken into account. The chooseshort effect appeared in all rats on and after 2-s RI [t-test: (RI 2-s): t(8)= 2.66, P B 0.05; (RI 3-s): t(8)= 3.65, PB 0.01; (RI 4-s): t(8)= 6.84, PB 0.001; (RI 5-s): t(8)= 4.04, P B 0.01]. The chooselong effect was again not significant. However, several significant effects emerged when the animals were split into two subgroups; one group included the five rats which presented more long than short correct responses and the other group included the remaining four animals which presented either an equal proportion of short and long correct responses or even a higher proportion of short correct responses. When statistical analyses were computed on the first group of five rats, a significant choose-long effect
P. Leblanc, M. Soffie´ / Beha6ioural Processes 56 (2001) 31–40
appeared whatever the three sessions [t(4) = 3.87, P B0.05] or only the first session [t(4) = 2.97, P B0.05] were taken into account. This effect never persists beyond five sessions. When the analyses were computed on the other group, no significant results appeared. The separation of experimental rats into two subgroups can not be explained by a procedural artefact such as a positional bias. As reported in the procedure, the position of the lever associated with the long duration was counterbalanced across the nine rats and the assignment of the levers remained random and did not differ into the two subgroups. Although the absence of procedural artefacts, the occurrence of choose-long responses can not be predicted by the performances of the animal during the former sessions. Indeed, never significant correlations were found between the preference for long response and either the ability to learn the duration discrimination (number of sessions to reach the criterion during training) or the extend or the persistence of the choose-short effect at 5-s RI, expressed, respectively by the differences between short or long correct responses and by the number of sessions to observe the disappearance of the choose-short preference.
2.3. Discussion The present experiment confirms our previous study (Leblanc and Soffie´ , 1999) having reported a choose-short effect in rats and confirms one of the above mentioned predictions, derived from the subjective-shortening model (Spetch and Wilkie, 1983). Indeed, the temporary occurrence of the choose-short effect was verified by the significant decrease in the mean difference between short and long correct responses and by the significant decrease in the choice ratios, observed when the first and the last blocks of three sessions at 5-s RI were compared. However, the choose-long effect was found only in five out of nine rats, when the RI was decreased to 0-s., whatever was the lever assignment. This effect disappeared more rapidly than the choose-short effect but persisted on and over the first session. No difference appeared in the
35
strength of the occurrence of both the chooseshort and choose-long effects, when only the first session after each change in the RI was taken into account. According to Spetch and Wilkie (1983), Spetch (1987), Grant et al. (1997), the choose-long effect occurred because a new reference memory, based on the working memory, would be formed at a fixed RI and because the working memory of the samples did not have been sufficiently foreshortened when the RI was decreased from 5 to 0-s. The absence of a choose-long effect in the remaining four rats, suggests that the reference memory would remain stable. It would be the content of the working memory which would become progressively close to the content of the reference memory and which would explain the disappearance of the choose-short effect. This supposed that the content of the working memory did not remain foreshortened, when a fixed RI was applied. The procedure used could be proposed to account for these contrasting results. Indeed, in our procedure, contrary to the classical stepwise delay procedure, 0-s training sessions were interposed between each series of the successive retention sessions in phase 1 (increases in the RI). It could be hypothesized that this modification of the procedure should have contributed to maintain a stable reference memory. In order to test this interpretation, a second experiment was carried out, using the classical stepwise delay procedure.
3. Experiment II In this experiment, young rats were trained in a classical stepwise delay procedure, without receiving 0-s training sessions between each RI series and with overtraining at each RI (Spetch and Wilkie, 1983).
3.1. Method 3.1.1. Subjects The subjects were young (6-month old) naı¨ve male Wistar (LOU/C) rats bred, reared and food deprived similarly as described in experiment I. At
36
P. Leblanc, M. Soffie´ / Beha6ioural Processes 56 (2001) 31–40
the beginning of the experiment, there were fourteen rats. As in experiment I, six rats which present a persistent choose- short effect beyond 20 sessions either at 4-s RI (N =4) or at 5-s RI (N = 2) were discarded and the data of only eight animals were analysed. The initial average body weight in grams9S.E.M. was 388.919 8.6 before the deprivation of food.
3.1.2. Apparatus The apparatus was the same as that used in Experiment I. 3.1.3. Procedure The procedure was similar to that used in Experiment I with two exceptions: (1) 0- s training sessions were not interposed between each RI series and (2) the animals were overtrained at each RI until the choose-short effect was disappeared. 3.1.3.1. Training. The training sessions were the same as that used in Experiment I 3.1.3.2. Retention testing. Following training sessions, the rats were submitted successively to phase 1 and to phase 2. Phase 1: Increases in the RI: During this phase, the RI was successively increased, for several consecutive sessions, to 1, 2, 3, 4 and 5-s RI. Within each RI series, the length of the interval remained constant on all trials. The rats were trained until they reached the criterion performance level for three sessions. Afterwards, they were submitted to phase 2. Phase 2: Decrease in the RI: During this phase, the RI was decreased from 5 to 0-s and the 0-s RI was maintained for three consecutive sessions, as in experiment I.
4. Results As shown in Fig. 2, during the first three individual sessions following each increase in the RI from 1 to 5-s, the choose-short effect was less consistent than in experiment I. It was only significant at three RIs and was not proportional to the retention delay. [t-test: (RI 1-s): t(7) = 2.73,
Fig. 2. Mean percentage of correct responses after short- and long-signal durations during the three first and the three last individual sessions after each change in the RI. The horizontal line indicates the significant threshold of correct responses (binomial test).
PB 0.05; (RI 3-s): t(7)=2.62, PB 0.05; (RI 4-s): t(7)= 2.9, PB0.05]. Moreover, when only the first session at each RI was taken into account, no choose-short effect appeared. With extended training at each RI, the chooseshort effect disappeared. To test the significance of the decrease of the choose-short effect at each RI, the first and the last blocks of three sessions were compared by t-tests. The tests were computed on the mean difference and also on the ‘choice ratios’. The decrease in the choose-short effect was expressed by a decrease in the mean difference [t-test: (RI 1-s): t(7)= − 4.48, P= 0.01; (RI 3-s): t(7)= − 3.57, P=0.01(RI 4-s): t(7)= − 3.17, P B0.05] and in the ‘choice ratio’ [t-test: (RI 1-s): t(7)= − 4.09, P= 0.01; (RI 3-s): t(7)= − 2.93, PB 0.05; (RI 4-s): t(7)= − 2.82, PB 0.05]. The mean number of sessions needed to observe a significant decrease at the different RIs were: 6.6 (RI 1-s), 4 (RI 2-s), 2.2 (RI 3-s), 6 (RI 4-s) and 9 (RI 5-s). As far as the three sessions following the decrease of the RI from 5 to 0-s were concerned, when the eight rats were taken into account, no significant choose-long effect appeared. The matching accuracy for both durations (i.e. the criterion level) reached during the last 5-s RI sessions was maintained in seven out of eight rats.
P. Leblanc, M. Soffie´ / Beha6ioural Processes 56 (2001) 31–40
Finally, a comparison of the percentage of correct responses between the two experiments did not show significant differences at each RI whatever the first or three first sessions were taken into account.
37
at the time of the duration sample and consequently, during extended training at a given RI, the rat could progressively learn to adopt a behavioural successful strategy which consisted in learning to maintain an orientation to the location of the correct lever over the RI.
5. Discussion 6. General discussion When training sessions were not interspersed between each RI series, as in the first experiment, the choose-short effect was less consistent. It appeared only when the three sessions at each RI were taken into account. These results are consistent with the procedure. It is obvious that an increase of 1-s between each RI was less disturbing than the increases of the RI up to 5-s imposed in experiment I. With extended training at each RI, the choose-short effect disappeared in the majority of rats, in which performances reached the criterion level. These results convey that the discrepancy between the working memory and the reference memory should have disappeared. Moreover, the hypothesis concerning the incidence of the procedure, proposed in the discussion of experiment I, was not verified. The absence of a significant choose-long effect and the recovery of accurate performances when the RI was decreased from 5 to 0-s suggested that the content of the reference memory would remain stable even when no training sessions were interspersed between each RI. Given that there was no evidence that the reference memory underwent a systematic change within the RI trials, the recovery of accurate performances with extended training during retention testing sessions should not be ascribed, as in pigeon’s studies, to a foreshortening of the reference memory. As proposed in experiment I, the disappearance of the chooseshort effect coming with the recovery of accurate performances could be, the consequence of an adaptation of the working memory which did not remain shortened. This interpretation is attractive, but unfortunately another non-memorial interpretation could explain the differences between rats and pigeons adaptations. In our experiment, the location of the correct response can be determined
The two experiments confirm, as already reported (Leblanc and Soffie´ , 1999; Soffie´ et al., 1999), that rats present a choose-short effect when a RI was applied between the offset of sample stimulus and the opportunity to respond. This effect was not often reported in rat studies which, unlike pigeon studies, often reported a chooselong effect (Roberts, 1982; Meck et al., 1984; Berz et al., 1992) and rarely a choose-short effect. Research with rats has shown that the apparition of the choose-short effect was largely influenced by procedural conditions. According to Santi et al. (1995, 1997), the rats evaluated the long duration of the sample as being ‘shorter’ compared to the objective duration in the context of a long ITI and with a change in lighting condition between the presentation of the sample and the ITI. Our experimental conditions (bimodal light/sound stimuli and 30 s-ITI), which were close to Santi’s conditions, could have contributed to promote the choose-short effect in our rats. Previous research, having reported a chooseshort-effect in rats, did not necessarily show a subjective-shortening of event duration. A forgetting which did not occur on time dimension was proposed by Church (1980) to explain the development of a choose-short bias. This interpretation appears unlikely to account for the development of the choose-short effect in our experiments, given that, contrary to Church’s experiment, a shift of the psychological function to the right and an increase of the point of subjective equality (PSE) was shown in our rats in similar experimental conditions (Leblanc and Soffie´ , 1999). According to the confusion/instructional failure hypothesis, proposed by Sherburne et al., (1998), if the ITI and the RI conditions match during test trials, the animal may not readily discriminate
38
P. Leblanc, M. Soffie´ / Beha6ioural Processes 56 (2001) 31–40
between them. Clearly spoken, if the animal had difficulty discriminating between the two intervals, it was uncertain what to do and when the two levers appeared after a RI, the experimental situation could be interpreted like a ‘no sample’ trial. Given that ‘no sample’ is closer to the short than to the long sample stimulus, the animal responded short. The observation of the behavioural strategy of our animals during the retention sessions did not support this confusion hypothesis. Indeed, systematic observations did disentangle two distinct strategies during RI and ITI: during RIs the rat remained in front of the box facing the levers and the feeding trough area, whereas during ITIs it walked around the whole skinner box. Finally, a procedural artefact such as positional response bias (Santi et al., 1995) could neither account for our results given that the chooseshort effect appeared in all rats and that the lever assignment was counterbalanced across our experimental rats. All these elements suggest that the choose-short effect, which appeared in rats in our experimental conditions, was not due to a procedural artefact and do not rule out a memorial factor such as a subjective-shortening. As far as the other predictions derived from the model are concerned, our results are not very conclusive. The predictions attempted to precise the interactive processes between working memory and reference memory involved when the choose-short effect disappeared. The temporary occurrence was verified in most but not in all rats. Given that the disappearance of the choose-short effect corresponded with the recovery of equivalent and significant correct performances for both short and long responses, the content of working memory would be similar to the content of reference memory. Pigeon’s experiments suggested that the disappearance of the choose-short effect was the consequence of the foreshortening of the duration stored in reference memory which became similar to the content of the duration stored in working memory during the current trial with RI. Accordingly, when a shorter RI was subsequently applied, the current sample duration appeared longer than the foreshortened reference memory
and the animal responded long. However, our data show that the majority of rats, instead of exhibiting a choose-long preference, maintained a good performance level when a shorter RI was applied. These results obtained in rats are consistent either with a different memory adaptation or with the development of an alternative behavioural strategy. The memory interpretation would suggest that the pivotal aspect between working memory and reference memory in rats was not the same as in pigeons and could be accounted for as follows: if rats maintained a stable reference memory during the RI sessions, a possibility to reduce the discrepancy between the two memories would be to adapt their working memory. In that case the rats, which had learned to evaluate the short- and the long-sample at the exact value even after a long 5-s RI, would not be disturbed during the subsequent shorter RI trials. During these shorter RI trials (0-s RI in our experiment) the working memory would not yet be involved and the information from timing processes would be directly compared to the durations stored in a stable reference memory. It is well known that rats are very spatial animals whereas pigeons are very visual animals. Accordingly, pigeons are more able than rats to learn a task, in which the correct choice stimulus did not remain in the same location on all trials and, which prevented the determination of the correct choice prior the RI. On the opposite, in rats studies, such as in our experiments, the correct lever associated with each duration remained in the same location and the rat, having the opportunity to know the correct choice prior the RI, can learn to maintain an orientation towards the location of the correct lever over the RI. Consequently, a non-memorial interpretation cannot be ruled out. However, the recuperation of accurate performances for both duration after the longest 4 and 5-s RI seemed not to be easy for the rat. Indeed, when the total of our experimental subjects were taken into account, 12 out of 29 rats did not succeed, even after 20 training sessions, to loose their choose-short preference either at 4 or at 5-s RI and in both experiments. Moreover, for rats which complete the experiments, a mean of, respectively 9.55 and 9 sessions were needed to
P. Leblanc, M. Soffie´ / Beha6ioural Processes 56 (2001) 31–40
recover the criterion performance level at 5-s RI. If rats were able to respond accurately, without the use of an effortful memory process (i.e., a working memory adaptation), but with the use of an alternative orientation strategy, how can we explain that 12 out of 29 rats did not succeed and that the remaining rats which complete the task did not recover the criterion performance level at 5-s RI more quickly? To summarize, the results of the studies expressed in this paper suggest that rats, as well as pigeons, are able to show a forgetting occurring on a time dimension and support the idea that the reference memory remain stable when the chooseshort effect occurred. However, there was no evidence, as in pigeons studies, that in rats the reference memory underwent systematic changes during retention testing sessions. Our results do not succeed to disentangle whether the recuperation of accurate performances during retention trials was the consequence of an adaptation of working memory or the consequence of the development of an alternative spatial strategy.
Acknowledgements We would like to thank Marie Bronchart and Marc-Henri Weyers for supervising the development of the programme of the experiment and data analyses and Christiane Van Vaerenbergh for her active collaboration in testing and data collection.
References Al Zahrani, S.S.A., Ho, M.Y., Velazquez Martinez, D.N., Lopez Cabrera, M., Bradshaw, C.M., Szabadi, E., 1996. Effect of destruction of the 5-hydroxytryptaminergic pathways on the acquisition of temporal discrimination and memory for duration in a delayed conditional discrimination task. Psychopharmacology 23, 103 –110. Berz, S., Ba¨ ttig, K., Welzl, H., 1992. The effects of anticholinergic drugs on delayed time discrimination performance in rats. Physiol. Behav. 51, 493 –499. Church, R.M., 1980. Short-term memory for time intervals. Learn. Motiv. 11, 208 –219. Fetterman, J.G., 1995. The psychophysics of remembered duration. Anim. Learn. Behav. 23 (1), 49 –62.
39
Fetterman, J.G., MacEwen, D., 1989. Short-term memory for responses: the choose-small effect. J. Exp. Anal. Behav. 52, 311 – 324. Grant, D.S., 1993. Coding processes in pigeons. In: Zentall, T.R. (Ed.), Animal cognition: a tribute to Donald A. Riley. Erlbaum, Hillsdale, NJ, pp. 193 – 196. Grant, D.S., Kelly, R., 1996. The role of minimum wait time and sample discriminability in the coding of event duration in pigeons. Learn. Motiv. 27, 243 – 259. Grant, D.S., Kelly, R., 1998. The effect of variable-delay training on coding of event duration in pigeons. Learn. Motiv. 29, 49 – 67. Grant, D.S., Spetch, M.L., 1991. Pigeons’ memory for event duration: differences between choice and successive matching tasks. Learn. Motiv. 22, 180 – 190. Grant, D.S., Spetch, M.L., 1993. Analogical and non-analogical coding of samples differing in duration in a choicematching task in pigeons. J. Exp. Psychol.: Anim. Behav. Process 19, 15 – 25. Grant, D.S., Spetch, M.L., 1994. Mediated transfer testing provides evidence for common coding of duration and line samples in many-to-one matching in pigeons. Anim. Learn. Behav. 22, 84 – 89. Grant, D.S., Spetch, M.L., Kelly, R., 1997. Pigeons’ coding of event duration in delayed matching to sample. In: Bradshaw, C.M., Szabadi, E. (Eds.), Time and Behavior: Psychological and Neurobehavioural Analyses. Elsevier, Amsterdam, pp. 217 – 251. Honig, W.K., 1978. Studies of working memory in the pigeon. In: Hulse, S.H., Fowler, H., Honig, W.K. (Eds.), Cognitive Processes in Animal Behavior. Erlbaum, Hillsdale, NJ, pp. 211 – 248. Kraemer, P.J., Mazmanian, D.S., Roberts, W.A., 1985. The choose-short effect in pigeon memory for stimulus duration: subjective-shortening versus coding models. Anim. Learn. Behav. 13, 349 – 354. Leblanc, P., Soffie´ , M., 1999. Effects of age on short-term memory for time in rats. Exp. Aging Res. 25, 267 – 284. Leblanc, P., Weyers, M.H., Soffie´ , M., 1996. Age-related differences for duration discrimination in rats. Physiol. Behav. 60 (2), 555 – 558. Meck, W.H., Church, R.M., Olton, D.S., 1984. Hippocampus, time and memory. Behav. Neurosci. 98, 3 – 22. Roberts, W.A., 1982. Cross-modal use of an internal clock. J. Exp.Psychol.: Anim. Behav. Process 8, 23 – 32. Roberts, W.A., Macuda, T., Brodbeck, D.R., 1995. Memory for number of light flashes in the pigeon. Anim. Learn. Behav. 23 (2), 182 – 188. Santi, A., Bridson, S., Ducharme, M.J., 1993. Memory codes for temporal and non-temporal samples in many to one matching by pigeons. Anim. Learn. Behav. 21, 120. Santi, A., Ducharme, M.J., Bridson, S., 1992. Differential outcome expectancies and memory for temporal and nontemporal stimuli in pigeons. Learn. Motiv. 23, 156 – 169. Santi, A., Standford, L., Coyle, J., 1997. Rats’memory for event duration: differential effects of delaying the discriminative choice cue as opposed to the opportunity 130 to execute the choice response. Behav.Proc. 40, 193 – 199.
40
P. Leblanc, M. Soffie´ / Beha6ioural Processes 56 (2001) 31–40
Santi, A., Weise, L., Kuiper, D., 1995. Memory for event duration in rats. Learn. Motiv. 26, 83 –100. Sherburne, L.M., Zentall, T.R., Kaiser, D.H., 1998. Timing in pigeons: the choose-short effect may result from pigeons’ ‘confusion’ between delay and intertrial intervals. Psychon. Bull. Rev. 5, 516 – 522. Soffie´ , M., Hahn, K., Terao, E., Eclancher, F., 1999. Behavioural and glial changes in old rats following environmental enrichment. Behav. Brain Res. 101, 37 –49. Spetch, M.L., 1987. Systematic errors in pigeons’ memory for event duration: interaction between training and test delay. Anim. Learn. Behav. 15, 1 –5. Spetch, M.L., Rusak, B., 1989. Pigeons’ memory for event duration: intertrial interval and delay effects. Anim. Learn. Behav. 17, 147 – 156. Spetch, M.L., Rusak, B., 1992a. Temporal context effects in pigeons’ memory for event duration. Learn. Motiv 23 (117), 144.
Spetch, M.L., Rusak, B., 1992b. Time present and time past. In: Honig, W.K., Fetterman, J.G. (Eds.), Cognitive Aspects of Stimulus Control. Erlbaum, Hillsdale, NJ, pp. 47 – 67. Spetch, M.L., Sinha, S.S., 1989. Proactive effects in pigeons’ memory for event duration: evidence against the coding model. J. Exp. Psychol.: Anim. Behav. Proc. 15, 347 – 357. Spetch, M.L., Treit, D., 1984. The effect of d-amphetamine on short-term memory for time in pigeons. Pharmacol. Biochem. Behav. 21, 663 – 666. Spetch, M.L., Wilkie, D.M., 1982. A systematic bias in pigeons’ memory for food and light duration. Behav. Anal. Lett. 2, 267 – 274. Spetch, M.L., Wilkie, D.M., 1983. Subjective-shortening: a model of pigeons’memory for event duration. J. Exp. Psychol.: Anim. Behav. Process. 9, 14 – 30. Wearden, J.K., Ferrara, A., 1993. Subjective shortening in humans’ memory for stimulus duration. Q. J. Exp. Psychol. 46 (2), 163 – 186.