Formal structure and pattern length in serial pattern learning by rats

When rats learn to anticipate a sequence of stimulus events, such as a serial pattern of different food quantities, they are sensitive to the rule-based formal structure relating the magnitude of successive stimuli. Earlier research has shown that if formal structure is simple (e.g., if a single “less than” rule relates the size of each successive quantity), patterns are learned faster than if formal structure is complex (e.g., if two or more rules such as “less than” and “greater than” relate successive pattern quantities). Two experiments tested the hypothesis that pattern length modulates the role of pattern complexity. We predicted that pattern length and pattern complexity interact in determining pattern difficulty. That is to say, long complex patterns should be learned more slowly than short complex patterns. However, long simple patterns should be learned faster than short simple patterns. In Experiment 1, rats ran a straight runway to receive repeated sequences of food quantities. The long-monotonic group received a formally simple 18-10-6-3-1-0 pattern, in which each number represents a quantity of food pellets. The long-nonmonotonic group received a formally complex 10-1-3-6-18-0 pattern. Similarly, the short-monotonic and short-nonmonotonic groups received 18-1-0 and 1-18-0 patterns. Pattern tracking—fast and slow running in anticipation of large and small quantities of food, respectively—was taken as an index of pattern learning. In Experiment 2, comparable patterns were used, but rats leverpressed in a discrete-trial procedure; response latencies measured pattern tracking. In both experiments, rats learned formally simple patterns faster than they did formally complex patterns. In Experiments 1 and 2, but less clearly in Experiment 2, the predicted interaction was obtained. The results support and generalize the idea that rats encode and use some representation of the formal rule structure of serial patterns as they learn them.     

Serial pattern learning by rats: Transfer of a formally defined stimulus relationship and the significance of nonreinforcement

Two experiments with rats tested independent predictions from cognitive theories of serial pattern learning. The animals learned to anticipate, as measured by running times in a straight alley, different quantities of food pellets organized into formally defined, five-element serial patterns. In Experiment 1, for some animals the patterns were all formally structured according to a monotonic “less than” relationship in which any quantity was always less than its predecessor. For others, no consistent formal rule was applied. Results of a transfer test with a new pattern showed positive transfer if the formal structure of the new pattern was identical to that used initially, but negative transfer if the new pattern was formally different. In Experiment 2, two groups learned the monotonic patterns 18-10-6-3-1 or 18-10-6-3-0 food pellets, while two others learned the nonmonotonic patterns 18-3-6-10-1 or 10-3-6-10-0 food pellets. We asked if the difference in value of the terminal element, 1 or 0 food pellets, would affect the facility with which the patterns were learned. The results showed that learning rate and the qualitative response to the elements of the monotonie and nonmonotonic patterns were independent of the value of the terminal element. Both experiments lend additional support to the utility of using cognitive models of human serial-pattern learning for an analysis of the sequential behavior of nonhuman animals.     

Extrapolation of serial stimulus patterns by rats

This experiment determined if rats could extrapolate a familiar serial sequence of diminishing food quantities by accurately anticipating a novel quantity added to the end of the sequence. In 13 days of training, rats ran in a straight runway to obtain quantities of food pellets presented in sequential order. A strongly monotonic group received repetitions of a formally simple pattern of 14-7-3-1 pellets of food, while weakly monotonic and nonmonotonic groups received formally more complex 14-5-5-1 and 14-3-7-1 patterns, respectively. In subsequent transfer, a 0-pellet quantity was added to each pattern, thus extending pattern length to five elements. Results of the very first pattern repetition containing the added 0-pellet element indicated that rats in the strongly monotonic condition, but not in the others, anticipated the reduced quantity before actually experiencing it. This result supports a cognitive, rule-learning hypothesis for serial learning by rats.     

An analysis of serial pattern learning by rats

When rats receive a sequence of rewards of different magnitudes for traversing a runway, they learn to “track” the sequence, showing anticipation of the forthcoming reward by appropriate running speed. There is disagreement as to whether this behavior depends on rats’ encoding and recalling a complete sequence of foregoing hedonic events or just the immediately preceding one. The present experiments showed that rats can remember more hedonic events than the most recent one. In Experiment 1, when exposed concurrently to the sequences 10-1-0 (pellets) and 0-1-10, they were faster on Run 3 of the increasing than of the decreasing sequence, a discrimination which cannot be made on the basis of the preceding (1-pellet) reward alone. Experiment 2 showed that this behavior reflects genuine anticipation of the Run 3 reward, not simultaneous contrast or other simple aftereffects of Runs 1 and 2. It is argued, however, that these results, together with related findings by Capaldi and Verry (1981), show merely that rats can recall a hedonic event other than the most recent one, not that a sequence of such events is fully recalled in order.     

Serial anticipation and pattern extrapolation in rats as a function of element discriminability

In Experiment 1, three groups of rats were trained on one of three serial patterns consisting of different numbers of .045-g food pellets, either the simple strong monotonie pattern 14-7-3-1-0 (Group SM) or one of two complex weak monotonie patterns, 14-5-5-1-0 (Group 5-5) or 14-9-1-1-0 (Group 1-1). Learning to anticipate the terminal 0-pellet element occurred faster in Group 1-1 than in Group SM, which in turn learned faster than Group 5-5. In Experiment 2, Groups SM, 5-5, and 1-1 were trained on the first four elements of the patterns experienced by their Experiment 1 counterparts and then were tested for their ability to extrapolate the series to include the addition of a 0-pellet element in the fifth position. Extrapolation behavior was better in Group 1-1 than in Group SM, which in turn performed better than Group 5-5. The results were shown to be consistent with a memory-discrimination model of serial learning and inconsistent with a rule-learning model based on pattern complexity.     

Element discriminability as a determinant of serial-pattern learning

Hulse and Dorsky found that rats were better able to track (run slowly to) 0 food pellets in a strongly monotonic (decreasing) serial pattern (14-7-3-1-0 food pellets) than in either a weakly monotonic one (14-5-5-1-0) or a nonmonotonic one (14-1-3-7-0). These findings were seen as incompatible with associative approaches based on animal experiments. Instead, they were taken to be consistent with cognitive theories of human behavior that relate pattern difficulty to formally defined structural complexity. In Experiment 1, tracking was found to be poorer with a strongly monotonie series (15-10-5-0) than with either of two weakly monotonic series (15-15-0-0 or 14-14-2-0), and in Experiment 2 a nonmonotonic series (1-29-0) produced better tracking than a strongly monotonic one (20-10-0). Although these results are not necessarily incompatible with the structural complexity view, they do suggest that “element discriminability” is a factor in serial-pattern learning. They are, therefore, compatible with a memory approach that views tracking as a form of discrimination learning.     

Spatiotemporal characteristics of serial CSs and their relation to search modes and response form

In four experiments, we examined how the spatiotemporal proximity to food of the two elements of a serial conditioned stimulus (CS) influenced the pattern of CS-directed versus food-site-directed behavior in rats. Experiment 1 showed that only temporal proximity affected responding when the serial CS consisted of two successive 4-sec presentations of either a spatially near or a spatially far lever (NN or FF). However, Experiment 2 showed that behavior depended markedly on whether rats received a near followed by a far lever (NF) or a far followed by a near lever (FN). Experiment 3 showed that the effects of Experiment 2 could be changed by increasing the duration of the second CS element, and Experiment 4 showed that these changes were not related to previous training. We concluded that behavior produced by the spatiotemporal qualities of the lever elements can be attributed to a mapping between the temporal qualities of the CS elements and an underlying sequence of search modes related to finding food.     

Preweanling and adult rats treat conditioned light-tone combinations differently

Ontogenetic differences in processing light-tone compounds were discovered in preweanling (17-day-old) and adult (60–80-day-old) rats. Suppression of general activity was used as an index of the magnitude of conditioned fear following a single training session in which a CS+ was paired with mild footshock. In Experiment 1, rats were trained on discriminations in which the CS? consisted of a light and the CS+ was either a tone alone (simple discrimination) or a light-tone compound (simultaneous feature-positive discrimination). Adults and preweanlings given each type of discrimination were then tested for fear of the CS? and a target stimulus (tone alone or light-tone compound). Adults in all groups displayed greater fear of the target than of the CS?. Preweanlings, however, discriminated the CS? from the target only when the target was the same as the original CS+. Experiment 2 revealed that age-related differences in conventional stimulus generalization is not a likely explanation for the pattern of results found in Experiment 1. Experiment 3 revealed age-related differences in expressed fear of a serial feature-positive discrimination; adults, but not preweanlings, showed greater fear of the compound than of the CS?. Alternative interpretations of the results from these experiments are discussed, and the general conclusion is that adults appear more inclined to process elements of a compound stimulus selectively, whereas preweanlings seem more likely to process the compound unselectively, with roughly equivalent processing of each element.     

Difficult serial anticipation learning in rats: Rule-encoding vs. memory

Rats, trained in a runway, were asked to anticipate, while running slowly, the last two events in repeating series of .045-g food pellets. The series were either weakly monotonic (14, 5, 5, 1, and then 0 pellets/run) or nonmonotonic (5-5-14-1-0). While the terminal 0-pellet event was better anticipated in the weakly monotonic series, the reverse was the case for the next-to-last 1-pellet event. These findings were expected from a memory-discrimination learning hypothesis of serial learning, which suggests that the memory of one event in a series can be used to signal the next event. However, the better anticipation of the 1-pellet event by the nonmonotonic group was inconsistent with the recently stated rule-encoding position of Hulse (1980). According to that view, difficult series of the sort employed in the present investigation are learned by encoding the rule structure of the series, with events in the series with the simple rule structure (the weakly monotonic series in this investigation) being better anticipated than events in the series with the complex rule structure.     

The nature of anticipation: An inter- and intraevent process

In these experiments, each rat received two different series of three runs each. The lone group in Experiment 1 received the series 10-0-10 and 10-0-0, where, for example, 10-0-10 means that the rat received three discrete runs (in a runway) that terminated in 10, 0, and 10 pellets, respectively. In Experiment 2, the series were 20-0-0 and 0-0-20 for one group and 20-0-20 and 0-0-0 for another. Of primary concern in both experiments was the rat’s anticipation, as measured by running speed, of 0 pellets on the middle, or second, run of each series. In each experiment, running speed to this 0-pellet event was independent of reinforcement magnitude on the first run of each series and was greater, the greater the reinforcement magnitude on the third run of each series. These results indicate that on the second run of each series the rat was anticipating not merely the 0 pellets associated with that run (intraevent anticipation), but also the reinforcement magnitude associated with the future, third run of each series (interevent anticipation). These results are shown to be consistent with an S-S cognitive view of anticipation and inconsistent with an S-R serial-chaining view of serial learning.     

Serial learning,interitem associations,phrasing cues,interference, overshadowing,chunking, memory,and extinction

Two experiments indicated that two approaches to serial learning are too extreme—the classical view that it consists only of interitem associations and various recent views that it involves no interitem associations. The novel assumption introduced here was that phrasing cues, normally conceptualized as merely segregating long series into smaller units or chunks, may also enter into associations with items, thereby reducing interitem interference and facilitating serial learning. It was found that one item could become a signal for another item, an interitem association, or be overshadowed by a phrasing cue, such as a brightness and temporal cue, also signaling that item. The items were .045-g pellets. Rats traversed a runway for items arranged in ordered series, 14-7-3-1-0 pellets (Experiment 1) or 10-2-0-10 (Experiment 2). Complete tracking of, for example, the 10-2-0-10 series would consist of fastest running to 10 pellets and slowest running to 0 pellets. In both investigations, the interitem association overshadowed was that between 0 pellets and the subsequent rewarded item, 0 → 14 (Experiment 1) or 0 → 10 (Experiment 2). Either repetitions of the 14-7-3-1-0 subpattern (Experiment 1) or merely the terminal 10-pellet item (Experiment 2) were phrased, both methods producing identical results. Overshadowing the 0-pellet item produced superior serial learning, more rapid extinction, and, in Experiment 1, considerable elevation of responding when the brightness phrasing cue was introduced in extinction, an effect said to be conceptually identical to spontaneous recovery and one demonstrating directly that phrasing cues are in reality overshadowing cues. It was suggested that many effects attributed to forgetting may be due to unrecognized overshadowing of memory cues by phrasing cues, giving rise to exaggerated estimates of forgetting.     

Covariation in conditioned response strength between stimuli trained in compound

Conditioned lick suppression by water-deprived rats was used to elaborate on recent evidence that the attenuated conditioned response elicited by an overshadowed stimulus may be enhanced by extinction of the overshadowing stimulus with which it had been trained in simultaneous compound. Using a modified serial stimulus arrangement in which a light coexisted with the last half of a tone that terminated with footshock, it was found in Experiment 1 that the tone overshadowed the light. Extinction of the tone-shock association resulted in a virtually complete recovery of the response to the overshadowed light. Using this serial overshadowing procedure, the possibility that the strength of a conditioned response to an element trained in compound covaries as a function of the strength of the response to the other element was tested in Experiment 2. Following overshadowing training similar to that of Experiment 1, independent reinforcement of the overshadowed light, that is, associative inflation, was found to have no deleterious effect on the response to the overshadowing tone. This suggests that the effects of postconditioning extinction and inflation of one element do not have symmetrical effects upon responding to the other element. The results of Experiment 2 were replicated in Experiment 3 using a simultaneous compound stimulus as opposed to the serial compound of the previous studies. These results are discussed in terms of various associative and cognitive models of learning and performance.     

What makes a landmark effective? Sex differences in a navigation task

In Experiment 1, two groups of female rats were trained in a triangular pool to find a hidden platform whose location was defined in terms of a single a landmark, a cylinder outside the pool. For one group, the landmark had only a single pattern (i.e., it looked the same when approached from any direction), while for the other, the landmark contained four different patterns (i.e., it looked different when approached from different directions). The first group learned to swim to the platform more rapidly than the second. Experiment 2 confirmed this difference when female rats were trained in a circular pool but found that male rats learned equally rapidly (and as rapidly as females trained with the single-pattern landmark) with both landmarks. This second finding was confirmed in Experiment 3. Finally, in Experiment 4a and 4b, male and female rats were trained either with the same, single-pattern landmark on all trials or with a different landmark each day. Males learned equally rapidly (and as rapidly as females trained with the unchanged landmark) whether the landmark changed or not. We conclude that male and female rats learn rather different things about the landmark that signals the location of the platform.     

Serial learning and transfer in rats: Effects of changes in stimulus-stimulus associations,pattern structure,and serial position information

Four groups of rats (n=6 per group) were trained in a runway on a serial learning task. Groups were treated identically in Phase 1, receiving two daily presentations of a five-element series consisting of decreasing numbers of .045-g food pellets over successive runs, for example, 14-7-3-1-0. All groups learned to anticipate, and run slowly to, the terminal 0-pellet element, behavior that has been attributed to learning of a less-than rule, stimulus-stimulus (S-S) associations, and knowledge of the serial position of items. In Phase 2, subjects were transferred to one of four test series: 20-14-7-3-0,20-7-3-14-0, 20-14-7-3-1-0, or 20-7-14-3-1-0. Anticipation was disrupted on the first two series, which maintained the integrity of serial position information and in the first case the less-than rule, but eliminated the terminal portion of the associative chain. Anticipation was unimpaired by transfer to either of the last two series. These series maintained the integrity of the terminal 3-1-0 portion of the associative chain but presented altered information about serial position, and in the last case also altered the less-than rule. The results, which supported the memory-discrimination model of rat serial learning, are discussed with reference to related transfer experiments in human serial learning.     

The development of visual expectations in the first year

The development of expectations was investigated by using the Visual Expectations Procedure. In Experiment 1, 128 infants aged 6-, 9-, and 12-months-old saw two 40-trial sequences of a videotaped mechanical toy appearing in various locations. The sequences represented an alternation pattern (i.e., ABAB) or a complex pivot pattern (i.e., ABCBABCB). In Experiment 2, 76 infants aged 4-, 8-, and 12-months-old saw either a left-right alternation or a top-bottom alternation. Reaction time improved and the percentage of anticipations increased between 6 and 9 months in Experiment 1 and between 4 and 8 months in Experiment 2 but not thereafter. Anticipations for the pivot sequence and for younger infants on both sequences were often incorrect (i.e., gaze shifts occurred before stimulus onset but were not directed toward the upcoming stimulus). We conclude that young infants have expectations that reflect some degree of general or procedural knowledge, but it is not clear that this behavior implies specific, articulated expectations about upcoming events.     

Conditioning of simultaneous and serial feature-positive discriminations

Three experiments with rat subjects examined the development of simultaneous and serial feature-positive discriminations in appetitive conditioning. In Experiment 1, reinforced presentations of a simultaneous light-tone compound were intermixed with nonreinforced presentations of either the light or the tone. The compound stimulus acquired conditioned behaviors of a form characteristic of the predictive feature alone; the element common to reinforced and nonreinforced trials did not evoke conditioned behavior. In Experiment 2, reinforced presentations of a serial light-trace-tone compound were intermixed with nonreinforced tone-alone presentations. The light feature stimulus acquired conditioned behaviors characteristic of visual CSs. The tone stimulus, common to reinforced and nonreinforced trials, evoked conditioned behaviors characteristic of auditory CSs, but only when preceded by the light. In Experiment 3, variations in the interval between the light and tone on reinforced trials had little effect on responding to the light CS but substantially altered the pattern of responding to the tone CS. These results suggested that simultaneous and serial feature-positive discriminations may be solved differently. Performance in simultaneous feature-positive discriminations may be determined solely by associations between the feature stimulus and the reinforcer, but performance in serial discriminations may also involve the acquisition of a conditional cue function to the feature.     

Artificial grammar learning in pigeons

In two experiments, we developed a new methodology for studying complex stimulus control by spatial sequences of letters generated by artificial grammars. An artificial grammar is a system of rules that defines which letter sequences or strings are “grammatical.” In Experiment 1, pigeons learned to respond differently to strings conforming to a grammar versus strings that were nongrammatical distortions. Several different criteria all suggested that performance was controlled both by some short chunks of strings shared between reinforced training strings and novel transfer strings and by more complex sequential regularities. In Experiment 2, pigeons quickly and accurately learned to respond differently to strings conforming to one or the other of two different artificial grammars. As in Experiment 1, performance was controlled both by some short chunks and by more complex sequential regularities. The results are interpreted in terms of family resemblance and pose new goals for theories of complex stimulus control.     

The role of trial tracking in rats’ working memory

The experiments reported in the present study tested whether decreasing intertrial intervals (ITIs) intensifies the disruptive effects of increasing retention intervals (RIs) in a delayed conditional discrimination by decreasing the animal’s trial tracking accuracy (Cohen & Armstrong, 1996; Cohen & Roberts, 1996). Rats responded on a fixed ratio (FR) 1 or fixed interval (FI) 10-sec reinforcement schedule at a second light or tone stimulus, S2, when the first light or tone stimulus, S1, had signaled an FI 10-sec or FR 1 schedule, respectively. RIs between S1 and S2 were increased from 3 to 24 sec and never exceeded ITIs that were reduced from 24 to 6 sec. For some rats, the trials were separated from each other by extending the lever at S1 and retracting it at the end of S2 (ITI lever-retracted group). For other, control rats, the lever remained extended throughout the session (lever-extended group, Experiment 1) or was extended and retracted with the onset and offset of each stimulus (RI/ITI lever-retracted group, Experiment 2). The rats under all trial conditions learned to delay leverpressing on the FI 10-sec schedule. Latency to begin leverpressing on the FI 10-sec schedule declined as RIs were increased, but this effect was attenuated in the ITI lever-retracted groups in both experiments, as would be predicted by thetrial tracking hypothesis. Decreasing ITIs from 24 to 6 sec intensified the disruptive effects of increasing RIs from 3 to 6 sec in the RI/ITI lever-retracted group (Experiment 2), as would be predicted by the trial tracking hypothesis.     

The representation of items in serial position

Four experiments with rats tested their ability to anticipate serial patterns made from elements of reward magnitudes (14, 7, 3, 1, or 0 food pellets). Anticipation was measured by the running time in a straight alley. Elements arranged in a monotonically descending pattern were more easily anticipated than were the same elements arranged in a nonmonotonic pattern. Better anticipation was also observed when training utilized four trials per day with short interrun intervals (10–15 sec), spent in the startbox of the runway, than when training utilized one trial per day with long interrun intervals (4–5 min), spent in the rat’s home cage. Anticipation of the monotonie sequence was also superior when training consisted of one trial per day with a short interrun interval relative to that observed with four trials per day and a long interrun interval. Following acquisition of anticipation of the monotonie sequence with a short interrun interval, transfer to the same sequence with a long interrun interval resulted in disruption of anticipation. Finally, anticipation of a well-learned monotonie sequence was not disrupted by replacement of individual rewarded elements in the sequence with a 0-pellet element. These experiments indicate that the duration between runs of a trial, but not that between trials or the number of trials per day, is important in the formation of serial expectancies. They also suggest that the rats come to represent the sequence as items in serial position.     

Processing of complex auditory stimuli (tunes) by rats and monkeys (Cebus apella)

The degree to which rats and monkeys base their discriminations of complex auditory stimuli (“tunes”) on frequency contours rather than on local features was investigated. In Experiment 1, groups of rats and monkeys trained with tunes as S+ and S? acquired a simple operant discrimination no faster than groups that received the same notes of each tune but in a new random order on each trial; neither did the groups differ on two transfer tests devised to detect learning of frequency contour in the tune-trained animals. Acquisition in the tune-trained and random-notes groups seemed to be based on the overall frequency difference between S+ and S?, which was about 1.5 octaves. In Experiment 2, S+ and S? were similar to each other with regard to overall frequency and individual notes, the most salient differentiating characteristic of the tunes being their tonal pattern. The tune-trained groups were clearly superior to the random-notes animals in acquisition, and an initial transfer test suggested that the former might have learned the discrimination on the basis of frequency contour. However, the detailed transfer tests of Experiment 3 strongly suggested that the tune-trained rats and monkeys based their discriminations primarily on local cues rather than on frequency contour. Based on the results of Experiment 4, the data of an earlier study that suggested frequency contour learning in monkeys and rats were reinterpreted in terms of control by local cues.