There's something about a pattern: Choice between pattern and random sequences in implicit learning.

Three experiments examined the preference for pattern versus random sequences. In all experiments the elements composing the sequences were visual images presented sequentially on a touchscreen. Reinforcement was randomly programmed on .16 of the element presentations for each type of trial. For pattern sequences the elements occurred in the same order and at the same location on each presentation of the sequence. For random sequences the elements could occur in any order on a given trial. The experiments were conducted in two phases. In the first phase, termed forced-choice, subjects, male Silver Kings, were given either a pattern sequence or a random sequence to work on in a given trial. Subjects received this first phase until performance on each type of sequence was equated. In the second phase, termed free-choice, subjects could choose which of the two sequences to work on in each trial. Results indicated that although performance was equated between the two types of sequences in the forced-choice phase, when given the choice subjects selected the pattern sequence on 70 percent of the trials. This finding held in Experiments 1 and 2 although there were procedural differences between these two experiments. In Experiment 3 the reinforcement probability for random sequences was increased to be 50 percent higher than for pattern sequence. In this arrangement subjects chose random sequences on nearly 83 percent of the free-choice trials, indicating that the preference for pattern sequences was not intractable. Results suggest that the preference for pattern sequences that was observed when reinforcement was equated between the two types of sequences may have been the result of the added information concerning forthcoming element presentations that was available from pattern sequences. (PsycInfo Database Record (c) 2023 APA, all rights reserved).

A swap methodology applied to implicit sequence learning.

Two experiments used a methodology in which elements in a serially presented sequence of 5 elements were randomly reinforced during training. To assess what was learned, elements were systematically swapped with each other during testing. The usual outcome measures in implicit sequence learning of this type are either a random test in which elements are disarrayed, or pairwise tests in which subjects choose between two elements. Each of these methods possesses shortcomings. The random test is a blunt measure, whereas pairwise tests disrupt the usual flow of elements in a serial sequence. Pairwise tests also present the problem of how to reinforce subjects during testing in a way that doesn't violate the delivery of reinforcement on a random basis. The swaps methodology is more targeted in that only two elements exchange ordinal positions. The procedure is at once more precise than the standard random test and is less disruptive to the flow of an implicit sequence than are pairwise tests. Results of both experiments indicated that the swaps manipulation was successful in disrupting latencies. Regression analyses indicated that subjects learned something about the ordinal position of elements. The separate analysis of symbolic distance versus spatial distance in Experiment 2 provided evidence, but only provisional evidence, of a symbolic distance effect in implicit learning. (PsycINFO Database Record

G and g: Two markers of a general cognitive ability, or none?

The search for general processes that underlie intelligence in nonhumans has followed two strategies: one that concerns observing differences between nonhuman species (G), the second that concerns observing individual differences within a nonhuman species (g). This commentary takes issue with both attempts to mark a general factor: Differential responding to contextual variables compromises the search for G, and the lack of predictive validity compromises g.

Implicit learning in cotton-top tamarins (Saguinus oedipus) and pigeons (Columba livia).

There is considerable interest in the conditions under which human subjects learn patterned information without explicit instructions to learn that information. This form of learning, termed implicit or incidental learning, can be approximated in nonhumans by exposing subjects to patterned information but delivering reinforcement randomly, thereby not requiring the subjects to learn the information in order to be reinforced. Following acquisition, nonhuman subjects are queried as to what they have learned about the patterned information. In the present experiment, we extended the study of implicit learning in nonhumans by comparing two species, cotton-top tamarins (Saguinus oedipus) and pigeons (Columba livia), on an implicit learning task that used an artificial grammar to generate the patterned elements for training. We equated the conditions of training and testing as much as possible between the two species. The results indicated that both species demonstrated approximately the same magnitude of implicit learning, judged both by a random test and by choice tests between pairs of training elements. This finding suggests that the ability to extract patterned information from situations in which such learning is not demanded is of longstanding origin.

Implicit chaining in cotton-top tamarins (Saguinus oedipus) with elements equated for probability of reinforcement.

Three experiments examined the implicit learning of sequences under conditions in which the elements comprising a sequence were equated in terms of reinforcement probability. In Experiment 1 cotton-top tamarins (Saguinus oedipus) experienced a five-element sequence displayed serially on a touch screen in which reinforcement probability was equated across elements at .16 per element. Tamarins demonstrated learning of this sequence with higher latencies during a random test as compared to baseline sequence training. In Experiments 2 and 3, manipulations of the procedure used in the first experiment were undertaken to rule out a confound owing to the fact that the elements in Experiment 1 bore different temporal relations to the intertrial interval (ITI), an inhibitory period. The results of Experiments 2 and 3 indicated that the implicit learning observed in Experiment 1 was not due to temporal proximity between some elements and the inhibitory ITI. The results taken together support two conclusion: First that tamarins engaged in sequence learning whether or not there was contingent reinforcement for learning the sequence, and second that this learning was not due to subtle differences in associative strength between the elements of the sequence.

Performance on patterned string problems by common marmosets (Callithrix jacchus).

This experiment examined the performance of common marmosets (Callithrix jacchus) on a series of patterned string problems to assess the marmosets' understanding of means-ends relationships. One marmoset, Jet, was exposed to a series of problems that were ordered in terms of perceived difficulty during two testings that were separated by 1 year. In the second testing, Jet received problems that had been used during the first testing along with three new problems. Each of the new problems was designed to be an exemplar of the type of problem that Jet had experienced difficulty with in the first testing. A second marmoset, Peaches, was tested on the same set of problems given to Jet in the second testing. Results indicated that the marmosets' performance on these problems fell into three categories. In one category, some problems were solved without evidence of trial-and-error learning. In a second category, there were problems in which the marmosets responded at chance levels initially but evidenced improvement as a function of extended testing. In a third category, some problems appeared to be virtually unsolvable even with extended testing. Taken together, these results indicate that the marmosets were able to learn the means-ends connection between pulling a string and obtaining food. This learning was best characterized as a trial-and-error process for some problem forms, while for others there appeared to be rapid learning that did not require extensive practice. The instances of rapid learning may be the result of the application of a simple spatial proximity rule in which the marmosets chose the string that was closest to an imaginary line drawn between the marmoset and the reinforcer.

Characteristics of implicit chaining in cotton-top tamarins (Saguinus oedipus).

In human cognition there has been considerable interest in observing the conditions under which subjects learn material without explicit instructions to learn. In the present experiments, we adapted this issue to nonhumans by asking what subjects learn in the absence of explicit reinforcement for correct responses. Two experiments examined the acquisition of sequence information by cotton-top tamarins (Saguinus oedipus) when such learning was not demanded by the experimental contingencies. An implicit chaining procedure was used in which visual stimuli were presented serially on a touchscreen. Subjects were required to touch one stimulus to advance to the next stimulus. Stimulus presentations followed a pattern, but learning the pattern was not necessary for reinforcement. In Experiment 1 the chain consisted of five different visual stimuli that were presented in the same order on each trial. Each stimulus could occur at any one of six touchscreen positions. In Experiment 2 the same visual element was presented serially in the same five locations on each trial, thereby allowing a behavioral pattern to be correlated with the visual pattern. In this experiment two new tests, a Wild-Card test and a Running-Start test, were used to assess what was learned in this procedure. Results from both experiments indicated that tamarins acquired more information from an implicit chain than was required by the contingencies of reinforcement. These results contribute to the developing literature on nonhuman analogs of implicit learning.

Implicit chaining in cotton-top tamarins (Saguinus oedipus).

In implicit learning, human subjects are exposed to patterned information, but they are not informed about the pattern. Typically, they demonstrate learning of that pattern, but little awareness of the experimental contingencies. In a nonhuman analog of this procedure, two cotton-top tamarins (Saguinus oedipus) were presented with a five-element chain that consisted of the same icon presented serially at different locations on a touchscreen. The tamarins had to touch the icon at each location to advance the chain and receive reinforcement at the end of the chain. One element of the chain was never differentially reinforced in the presence of another element, as is typically done in transitive inference and serial chaining studies. Following training, the tamarins were tested for their knowledge of the chain using pairwise tests that are common in transitive inference and serial chaining experiments, and a random test, common in some types of implicit learning, in which the sequence of elements was randomized. The results of both tests revealed that the tamarins appreciated the ordinal position of the elements composing the chain, although reinforcement had not been dependent on that knowledge.

Further evidence that mice learn a win-shift but not a win-stay contingency under water-escape motivation.

Previous research (C. Locurto, C. Emidy, & S. Hannan, 2002) indicated that mice quickly learned a water-escape task under a win-shift contingency but did not exceed chance-level performance under a win-stay contingency. We examined the robustness of this conclusion in two experiments by varying procedural and temporal aspects of that earlier experiment. Results of both experiments indicated that the preference for win-shift learning in mice under water-escape motivation could not be attributed to procedural or design features of that earlier study and were independent of the influence of intertrial interval, normally a variable that produces strong effects on learning. In neither experiment did subjects exposed to a win-stay contingency perform at above-chance levels.

Mice (Mus musculus) learn a win-shift but not a win-stay contingency under water escape motivation.

Twenty mice (Mus musculus), the second filial generation offspring from a C57BL/6 and DBA/2J cross, received spatial win-shift and win-stay water escape training within a mixed design in which all mice received both types of training. Acquisition under win-shift was superior to win-stay with respect to errorless trials and latencies regardless of the order in which the procedures were experienced. Win-stay responding did not exceed chance levels during any training phase. These data contradict the claim that win-stay training is the more easily acquired of the 2 acquisition strategies under aversive motivation.