Higher Order Chunking in Serial Pattern Learning by Rats in the T-Maze

Rats were repeatedly exposed to two- and three-trial series consisting of different reinforced (R) and nonreinforced (N) trials in a fixed order on extended visually distinct runways in a T-maze. When initially presented with the same sequence, RR or RN in Experiment 1 and RRN or RNR in Experiment 2, on each series in a session (Separate Presentations), rats developed slower running speeds or bypassed the runway in returning to the start chamber on N more than on R trials. These serial response patterns occurred whether rats experienced each sequence on a specific runway (Associated group) or equally on different runways over sessions (Nonassociated group). When sequences were intermixed within each session, only the Associated group maintained its serial pattern responses to both sequences while the Nonassociated group only did so within the RRN sequence. When allowed to choose between runways over trials, the Associated group tended to select the runway associated with the RR sequence on the first two free-choice trials while rats in the Nonassociated group tended to alternate their choices of runways. Switching the relevancy of the runway context in the second experiment caused rats in each group to react like rats in the other group on forced-choice trials. In terms of Capaldi's (1992) model of levels of chunking in serial pattern learning, our findings indicate that rats learned to anticipate at least second trial outcomes by using higher order list or contextually cued chunking and also to anticipate third trial outcomes by using lower order series chunking.     

The Equation of the Galvanic Skin Reflex Curve: I. The Dynamics of Reaction in Relation to Excitation-Background

Previous studies of double alternation learning have failed to demonstrate learning of extended double alternation sequences by nonprimate Ss. The present study investigates whether such learning can be induced by manipulations of the training procedure. Pigeons (N = 6) were trained on a double alternation procedure with continuous trial presentations. A new double alternation sequence began after each error so that the length of a sequence was a function of the accuracy of performance. All Ss learned the double alternation pattern for the first four positions of the sequence, but none learned consistently to emit longer sequences. The degree of learning was a function of several procedural variables, particularly the use of an appropriate correction procedure.     

Comments on Shimp's (1983) double dissociation between knowledge and tacit knowledge

Shimp (1983) found in five pigeons “double dissociation” between the distribution of pairs of long and short reinforced interresponse times and “self-reports,” obtained by symbolic matching to sample, concerning these interresponse times. This result can be explained by assuming that the birds used a fixed temporal interval as matching criterion, independently of which pattern of interresponse times was reinforced.     

Single and double alternation learning in rats: The role of set size and correction

In many experiments, rats have evidenced extreme difficulty mastering alternation patterns. In three experiments, we explored rats’ ability to learn double alternation patterns and possible reasons behind their past difficulties with such patterns. In Experiment 1, rats learned single and double alternation patterns. In the second and third experiment, we explored the role of correction and interference from stimulus set size on rats’ learning of double alternation patterns. Our results showed that rats easily learned both single and double alternation patterns and that stimulus set size had no noticeable effect on learning double alternation patterns. Findings additionally evidenced that correction played an important role in rats’ learning double alternation patterns.     

Effects of a DRL contingency added to a fixed-interval reinforcement schedule

Following 30 days of reinforcement for the bar press response of two white rats on 30-sec fixed-interval (FI), a DRL component was added so that a minimal interresponse time (IRT) for the reinforced response, in addition to the FI variable, was necessary for reinforcement. Marked control over response rate by the superimposed DRL requirement was demonstrated by an inverse hyperbolic function as the DRL component was increased from 1 to 24 sec within the constant 30-sec FI interval. Interresponse time and post-reinforcement (post-S^R) “break” distributions taken at one experimental point (DRL = 24 sec) suggested that a more precise temporal discrimination was initiated by an S^R than by a response, since the relative frequency of a sequence of two reinforced responses appeared greater than that of a sequence of a non-reinforced response followed by a reinforced one. This latter finding was confirmed with new animals in a follow-up experiment employing a conventional 24-sec DRL schedule.     

Comparing the effects of two correction procedures on human acquisition of sequential behavior patterns

Thirty-one college undergraduates learned to touch abstract stimuli on a computer screen in arbitrarily designated “correct” sequential orders. Four sets of seven stimuli were used; the stimuli were arrayed horizontally on the screen in random sequences. A correct response (i.e., touching first the stimulus designated as first) resulted in that stimulus appearing near the top of the screen in its correct sequential position (left to right), and remaining there until the end of the trial. Incorrect responses (i.e., touching a stimulus out of sequence) terminated the trial. New trials displayed either the same sequence as the one on which an error had occurred (same-order correction procedure), or a new random sequence (new-order correction procedure). Whenever all responses occurred in the correct sequence, the next trial displayed a new random sequence. Each phase ended when five consecutive correct response sequences occurred. Initially, the same-order correction procedure increased control by the position as well as by the shape of the stimuli; also, it produced more errors, more total trials, more trials to mastery, and more individual patterns of reacquisition than were produced by the new-order procedure.     

Integration and representation in rats' serial pattern learning in the T-maze

Rats were exposed to three-trial series consisting of reinforced (R) trials and one nonreinforced (N) trial in a fixed order, RRN and RNR (Experiments 1 and 2) or NRR and RRN (Experiment 3), on extended visually distinct runways in a T-maze. When initially presented with the same sequence on each series in a session (separate presentations) with the same runway on all trials within a series (Experiments 1 and 3), all the rats developed slower running speeds on N than on R trials. When a runway was sometimes changed between the first and next two trials during separate presentations training (Experiment 2) or both sequences were later intermixed within each session in each experiment, only rats exposed to each sequence on a specific runway maintained these serial running patterns. Rats displayed serial running patterns on a test RNN sequence similar to that on the RNR sequence (Experiment 2), as would be predicted by an intertrial association model of serial pattern learning (Capaldi & Molina, 1979), but responded on test RRR and NRN sequences (Experiment 3) as would be predicted by an ordinal-trial-tag/intratrial association model (Burns, Wiley, & Payne, 1986). Results from test series of free-choice trials in Experiments 1 and 2 failed to support a prediction of the intratrial association model that these rats would integrate RRN and RNR sequences. Rather than always selecting a baited runway on both the second and the third free-choice trials, the rats only selected a baited runway on the third trial on the basis of their choice on the second trial, as would be predicted by the intertrial association model. Only after experiencing all possible outcome sequences during forced-choice training in Experiment 3 did these rats predominantly select a baited runway on every free-choice trial.     

Rats’ anticipation of current and future trial outcomes in the ordered RNR/RNN serial pattern task

In the ordered RNR/RNN serial pattern task, rats often reduce their running speeds on trial 2 less within the RNR than within the RNN series. Initially, investigators (Capaldi, 1985; Capaldi et al., 1983) considered this trial 2 differential speed effect evidence for rats’ anticipation of inter-trial outcomes within each series. Later findings, however, suggest that this effect reflects some generalization of the ordinal position of trial 3 (Burns et al., 1986) or its similar runway cues during trial 2 (Capaldi et al., 1999). To test these generalization hypotheses, we made trial 2 more distinct from trial 3 in each series by forcing rats to alternate runways in a T-maze only on the last trial rather than on trial 2 in each series in Experiment 1, or by forcing rats to alternate runways between trials rather than to run down the same runway on all trials within each series in Experiment 2. Although enhancing the distinctiveness between these trials reduced the trial 2 differential speed effect, extensive training failed to eliminate it. Therefore, this residual difference between trial 2 speeds could reflect rats’ anticipation of trial 3 outcomes during trial 2 as originally proposed by Capaldi (1985) Experiment 3 was designed to determine whether we could enhance rats’ final trial outcome expectancies during trial 2 by making different trial 2 choices distinctive cues for each trial 3 outcome. The trial 2 speed effect was greater when rats were forced to alternate over all trials only within one of the series than when they were sometimes forced to do so in either series. Post-training probe tests revealed that both series position and the relevant within-series runway events contributed to this enhanced anticipation of trial 3 outcomes.     

Early stimulation and patterned responding in rats

This experiment was designed to determine the extent to which performance differences on learning tasks arising from infantile handling reflect learning or emotionality differences. Handled and nonhandled rats were trained and tested in a controlled operant situation where a response on every other trial yielded reinforcement. A key light was turned on for 10 sec during each trial and was off during the 10-sec intertrial interval. The reinforced (S+) trials alternated in a regular sequence with nonreinforced (S–) trials. The pattern of S+ and S– trials recycled continuously throughout the session of 160 trials. The handled rats emitted more responses during both S+ and S– trials than did the nonhandled rats. However, when they were compared on the percentage of emitted responses that were reinforced, there was no significant differences between the two groups. These results were interpreted to suggest that infantile handling does not directly facilitate associative learning in rats.This research was aided by a grant from the National Research Council of Canada, 67-0247. I thank Linda Easton and Beverly Clark for their help in collecting and analyzing the data.     

Spontaneous alternation and pattern of reinforcement

In Experiment I, rats were given one pair of trials per day in a T-maze; group RN were rewarded on the first trial only, NR were rewarded on the second trial only, while RR were rewarded on both trials. Alternation was approximately the same for all groups, and there was little decline in alternation over 28 days. There was, however, a difference in the choice behaviour on trials when rats failed to alternate, NR and RR, but not RN, choosing according to a position preference. These results are incompatible with several earlier theories of spontaneous alternation, and the hypothesis was put forward that alternation is initially part of an exploratory tendency, but that, when the maze becomes explored, the exploratory tendency dies down, and alternation only continues when rats have been reinforced (primarily or secondarily) for alternating. This hypothesis was tested in a second experiment, similar to the first but in which one group of rats was not rewarded on either trial; as predicted, this group showed significantly more decline in alternation than rewarded rats.     

Transposition of flower height by bumble bee foragers (Bombus impatiens)

Honeybees learn visual characteristics of reinforced and non-reinforced flowers in differential conditioning experiments (i.e., experiments that require subjects to choose between a reinforced and non-reinforced flower). In this study bumble bee foragers (Bombus impatiens) were trained in a transposition paradigm to determine if the relational properties of flowers also influence choice behavior. Subjects in one group (group A) were trained with repeated choices between a medium-height flower replete with sucrose solution and an empty tall flower until the medium-height flower was sampled preferentially in five consecutive trials. A second group (group B) was trained on the medium height flower alone for five trials. In a single test trial subjects were given a choice between the medium-height flower and a short flower, each filled with water. A control group showed no preference in this test. Group B subjects showed a significant preference for the medium-height flower and group A subjects that exhibited flower constancy (i.e., sampled only the medium-height flower in training trials) showed an identical pattern of choice. Subjects in group A that sampled both flowers during training, in contrast, transposed flower height and preferred the short flower. These results suggest that the choice behavior of bumble bee foragers is influenced by relational and absolute properties of flowers. The flower characteristics learned by foragers appear to depend on the difficulty of the discrimination problem and the context in which flowers are sampled. Patches of flowers limited in phenotypic variability may produce simple associative learning and flower constancy by foragers. Received: 17 September 1999 / Accepted after revision: 22 June 2000     

Short-term memory in the pigeon: the previously reinforced response

Eighteen pigeons served in a discrete-trials short-term memory experiment in which the reinforcement probability for a peck on one of two keys depended on the response reinforced on the previous trial: either the probability of reinforcement on a trial was 0.8 for the same response reinforced on the previous trial and was 0.2 for the other response (Group A), or, it was 0 or 0.2 for the same response and 1.0 or 0.8 for the other response (Group B). A correction procedure ensured that over all trials reinforcement was distributed equally across the left and right keys. The optimal strategy was either a winstay, lose-shift strategy (Group A) or a win-shift, lose-stay strategy (Group B). The retention interval, that is the intertrial interval, was varied. The average probability of choosing the optimal alternative reinforced 80% of the time was 0.96, 0.84, and 0.74 after delays of 2.5, 4.0, and 6.0 sec, respectively for Group A, and was 0.87, 0.81, and 0.55 after delays of 2.5, 4.0, and 6.0 sec, respectively, for Group B. This outcome is consistent with the view that behavior approximated the optimal response strategy but only to an extent permitted by a subject's short-term memory for the cue correlated with reinforcement, that is, its own most-recently reinforced response. More generally, this result is consistent with “molecular” analyses of operant behavior, but is inconsistent with traditional “molar” analyses holding that fundamental controlling relations may be discovered by routinely averaging over different local reinforcement contingencies. In the present experiment, the molar results were byproducts of local reinforcement contingencies involving an organism's own recent behavior.     

Effects of anxiolytic drugs on the partial reinforcement extinction effect in runway and skinner box

The partial reinforcement extinction effect (PREE) in the runway is reduced by anxiolytics in a non-state-dependent manner when a 24 h inter-trial interval is used, but there is some doubt as to the nature of the drug effects when shorter intervals are used. Experiment 1 repeated a study by Gray (1969), in which ambiguous results were obtained using eight trials/day. It demonstrated that the anxiolytic barbiturate, sodium amylobarbitone, given both in acquisition and extinction does not reduce the PREE. It confirmed Gray's observation that the PREE is abolished if the drug is given in acquisition but not in extinction. This suggests that a 24 h inter-trial interval is one critical factor in non-state-dependent reduction of the PREE. Experiments 2 and 3 tested the effects of the anxiolytic benzodiazepine, chlordiazepoxide, on the PREE with a 24 h inter-trial interval in the Skinner box. The basic task was a single FR5 sequence terminating in delivery of 10 (Experiment 2) or 20 (Experiment 3) reward pellets each day. There were 10 acquisition trials and partially reinforced rats received either three (Experiment 2) or four (Experiment 3) non-rewarded trials. The drug abolished the PREE in Experiment 2 and effectively reversed it in Experiment 3. These results confirm previous work with this drug in the runway (Feldon and Gray, 1981) and extend them to a very different experimental situation. These results support the idea that the PREE depends on different processes with different acquisition parameters; and that when a 24 h inter-trial interval is used the PREE is largely produced by some general process, probably the counterconditioning of conditioned frustration, which is sensitive to anxiolytic drugs. They also demonstrate very clearly the paradoxical effects of the anxiolytic drugs when given in both acquisition and extinction: they generally increase resistance to extinction in continuously reinforced animals, but block the increase in resistance (the PREE) produced by behavioural schedules.     

Repetition and alternation in rats

An experiment is reported in which rats were forced equally often to each arm of a T-maze, and, following each forced trial, were given a free trial. Group R were rewarded for repeating (i.e. choosing the side to which they had just been forced), group A for alternating. All rats began by alternating, but group R eventually learned to repeat. Various delays between free and forced trials were introduced, and the forgetting curve was found to be similar for the two groups. It is argued that these results show: (i) The decline in alternation with delay found in experiments on spontaneous and rewarded alternation is due to loss of information rather than a decline in the alternation tendency. (ii) The trace involved is sufficiently general to be regarded as a memory trace. The fact that rats can learn to repeat is inconsistent with the accounts of the memory trace suggested by the theories of Deutsch and Walker.     

Response patterning in goldfish: single alternation in a Pavlovian aversive paradigm

Single alternation behavior was studied in a Pavlovian aversive shuttle-response situation using goldfish. Independent groups of fish were given either a fixed interval between alternating reinforced (R) and nonreinforced (N) trials or were given differential temporal information between trials. Although all groups exhibited higher response probabilities on reinforced than on nonreinforced trials, goldfish receiving a short interval (10 s) following R trials and a long interval (60 s) following N trials (R10N60) demonstrated significantly superior alternation performance compared with subjects receiving a fixed intertrial interval (R35N35) or a long interval following R trials and a short interval following N trials (R60N10). The alternation performance exhibited by the R10N60 group was shown to be equivalent to that of subjects receiving standard color discrimination training. However, the alternation performance in group R35N35 showed a great deal of within-subject variability and raises questions concerning the elusive nature of alternation using Pavlovian conditioning procedures.     

The effect of blocking inter-trial interval on sequential effects in absolute identification

Sequential effects are ubiquitous in decision-making, but no more than in the absolute identification task where participants must identify stimuli from a set of items that vary on a single dimension. A number of competing explanations for these sequential effects have been proposed, and recently Matthews and Stewart [(2009a). The effect of inter-stimulus interval on sequential effects in absolute identification. The Quarterly Journal of Experimental Psychology, 62, 2014–2029] showed that manipulations of the time between decisions is useful in discriminating between these accounts. We use a Bayesian hierarchical regression model to show that inter-trial interval has an influence on behaviour when it varies across different blocks of trials, but not when it varies from trial to trial. We discuss the implications of both our and Matthews and Stewart's results on the effect of inter-trial interval for theories of sequential effects.     

Chlordiazepoxide and resistance to punishment

Rats were given intermittent electric foot-shock during food-rewarded alley training. In the test phase, food and shock were given on every trial. These animals persisted in running down the alley in the test phase compared to those without prior shock experience. The effects of chlordiazepoxide (CDP) on this learned resistance to punishment were examined using a long and short interval between trials. It was found that CDP abolished the effect at a long inter-trial interval, but left it unaltered if the interval was short. The results match those found previously with an analogous effect using non-reward. It is suggested that the effects of punishment and non-reward may be mediated by a common process, and that the benzodiazepines may act on this process.     

Estrogen Improves Performance of Reinforced T-Maze Alternation and Prevents the Amnestic Effects of Scopolamine Administered Systemically or Intrahippocampally

In a previous study, administration of high doses of estradiol benzoate (100 μg/kg for 3 days im) to ovariectomized Long-Evans rats counteracted impairments of reinforced T-maze alternation induced by systemic administration of scopolamine, a muscarinic receptor blocker. In the current study, daily administration of lower doses of estradiol benzoate (5 μg/kg for 3 weeks sc) increased the number of correct reinforced alternations during T-maze acquisition in ovariectomized rats compared to oil-treated controls and prevented impairments of reinforced alternation induced by injection of scopolamine hydrobromide (0.2 mg/kg ip). Furthermore, scopolamine (20 μg) delivered bilaterally to the dorsal hippocampus reduced reinforced T-maze alternation in ovariectomized rats previously trained to complete this task while daily treatment with estradiol benzoate (5 μg/kg sc) for 1 week prior to scopolamine infusion counteracted this impairment. In summary, physiological levels of estrogen improved performance during acquisition of reinforced T-maze alternation and prevented impairments induced by scopolamine administered systemically or intrahippocampally.     

Sequential patterns in post-reinforcement pauses on fixed-interval schedules of food

Responding by one pigeon was reinforced with food on fixed-interval schedules of 30, 60, and 300 sec duration. A second pigeon was studied under fixed-interval durations of 60 and 300 sec. For both birds, the average post-reinforcement pause was one-half the duration of the fixed interval. Autocorrelation coefficients revealed first-order sequential dependencies in series of post-reinforcement pauses. On the 300-sec fixed-interval schedule, successive post-reinforcement tended to alternate between long and short durations. At the shorter fixed-interval durations there was less evidence of alternation sequences. A second experiment was conducted to determine if the time intervals between the first response after reinforcement and the next reinforcement (the work periods) were responsible for the alternation patterns in the series of post-reinforcement pauses. To evaluate the role of the work period, several procedures were used to modify the work period from that obtained on the fixed-interval 300-sec schedule. Adding a schedule to the fixed-interval schedule that set the minimum amount of time that could elapse between the first response after reinforcement and the next reinforcement eliminated the alternation pattern. Control schedules indicated that the elimination of alternation patterns resulted from constraints on the work period per se and not from confounded changes in the interreinforcement intervals.     

Pigeons'' spatial memory: II. Acquisition of delayed matching of key location and transfer to new locations

Five hungry pigeons first received delayed matching of key location training. Trials began with a “ready” stimulus (brief operation of the grain feeder). Then one (randomly chosen) of a set of four keys from a three-by-three matrix was lit briefly as the sample. After a short delay (retention interval), the sample key was lit again along with one of the other eight keys. A peck at the key that had served as the sample produced grain reinforcement, whereas a peck to the other key produced only the intertrial interval. After delayed matching of key location was learned, the remaining five key locations were introduced as samples. Four of the five birds performed at considerably above-chance levels on the novel sample trials during the first as well as subsequent sessions. These results suggest that pigeons sometimes learn the single rule—“choose the location that matches the sample.” The relevance of these results to the issue of whether pigeons learn a generalized matching rule (i.e., a concept of “sameness”) is discussed.