Evidence for a conceptual account of same-different discrimination learning in the pigeon

From Concrete Examples to Abstract Relations: The Rostrolateral Prefrontal Cortex Integrates Novel Examples into Relational Categories

The ability to form relational categories for objects that share few features in common is a hallmark of human cognition. For example, anything that can play a preventative role, from a boulder to poverty, can be a "barrier." However, neurobiological research has focused solely on how people acquire categories defined by features. The present functional magnetic resonance imaging study examines how relational and feature-based category learning compare in well-matched learning tasks. Using a computational model-based approach, we observed a cluster in left rostrolateral prefrontal cortex (rlPFC) that tracked quantitative predictions for the representational distance between test and training examples during relational categorization. Contrastingly, medial and dorsal PFC exhibited graded activation that tracked decision evidence during both feature-based and relational categorization. The results suggest that rlPFC computes an alignment signal that is critical for integrating novel examples during relational categorization whereas other PFC regions support more general decision functions.

Detection and discrimination of complex sounds by pigeons (Columba livia)

Auditory scene analysis is the process by which sounds are separated and identified from each other and from the background to make functional auditory objects. One challenge in making these psychological units is that complex sounds often continuously differ in composition over their duration. Here we examined the acoustic basis of complex sound processing in four pigeons by evaluating their performance in an ongoing same/different (S/D) task. This provided an opportunity to investigate avian auditory processing in a non-vocal learning, non-songbird. These pigeons were already successfully discriminating 18.5 s sequences of all different 1.5 s sounds (ABCD…) from sequences of one sound repeating (AAAA…, BBBB…, etc.) in a go/no-go procedure. The stimuli for these same/different sequences consisted of 504 tonal sounds (36 chromatic notes×14 different instruments), 36 pure tones, and 72 complex sounds. Not all of these sounds were equally effective in supporting S/D discrimination. As identified by a stepwise regression modeling of ten acoustic properties, tonal and complex sounds with intermediate levels of acoustic content tended to support better discrimination. The results suggest that pigeons have the auditory and cognitive capabilities to recognize and group continuously changing sound elements into larger functional units that can serve to differentiate long sequences of same and different sounds.

No evidence for feature binding by pigeons in a change detection task

We trained pigeons to respond to one key when two consecutive displays were the same as one another (no-change trial) and to respond to another key when the two displays were different from one another (change trial; change detection task). Change-trial displays were distinguished by a change in all three features (color, orientation, and location) of all four items presented in the display. Pigeons learned this change-no change discrimination to high levels of accuracy. In Experiments 1 and 2, we compared replace trials in which one or two features were replaced by novel features to switch trials in which the features were exchanged among the objects. Pigeons reported both replace and switch trials as "no-change" trials. In contrast, adult humans in Experiment 3 reported both types of trials as "change" trials and showed robust evidence for feature binding. In Experiment 4, we manipulated the total number of objects in the display and the number of objects that underwent change. Unlike people, pigeons showed strong control by the number of feature changes in the second display; pigeons' failure to exhibit feature binding may therefore be attributed to their failure to attend to items in the displays as integral objects.

Avian cognition: examples of sophisticated capabilities in space and song

Although birds have traditionally and colloquially been considered less cognitively complex than mammals, and especially primates, more recent research has consistently refuted these assumptions. We argue that the impressive abilities of birds to navigate and communicate require considerable information-processing capabilities. These capacities include collecting, organizing, and selecting from a wide variety of navigational cues to orient toward and find a goal location in the spatial domain, and utilizing open-ended categorization and possibly even abstract reasoning to discriminate species-specific acoustic features of songs and calls. Furthermore, these abilities may be present across many avian species, providing evidence for domain-general cognitive facilities. We provide examples of processes in spatial learning and communication in birds, and locate them within the general literature, as evidence that the term 'bird-brain' should not be considered a pejorative.

Two-item same/different discrimination in rhesus monkeys (Macaca mulatta)

Almost all nonhuman animals can recognize when one item is the same as another item. It is less clear whether nonhuman animals possess abstract concepts of "same" and "different" that can be divorced from perceptual similarity. Pigeons and monkeys show inconsistent performance, and often surprising difficulty, in laboratory tests of same/different learning that involve only two items. Previous results from tests using multi-item arrays suggest that nonhumans compute sameness along a continuous scale of perceptual variability, which would explain the difficulty of making two-item same/different judgments. Here, we provide evidence that rhesus monkeys can learn a two-item same/different discrimination similar to those on which monkeys and pigeons have previously failed. Monkeys' performance transferred to novel stimuli and was not affected by perceptual variations in stimulus size, rotation, view, or luminance. Success without the use of multi-item arrays, and the lack of effect of perceptual variability, suggests a computation of sameness that is more categorical, and perhaps more abstract, than previously thought.

Endpoint distinctiveness facilitates analogical mapping in pigeons

Analogical thinking necessitates mapping shared relations across two separate domains. We investigated whether pigeons could learn faster with ordinal mapping of relations across two physical dimensions (circle size & choice spatial position) relative to random mapping of these relations. Pigeons were trained to relate six circular samples of different sizes to horizontally positioned choice locations in a six alternative matching-to-sample task. Three pigeons were trained in a mapped condition in which circle size mapped directly onto choice spatial position. Three other pigeons were trained in a random condition in which the relations between size and choice position were arbitrarily assigned. The mapped group showed an advantage over the random group in acquiring this task. In a subsequent second phase, relations between the dimensions were ordinally reversed for the mapped group and re-randomized for the random group. There was no difference in how quickly matching accuracy re-emerged in the two groups, although the mapped group eventually performed more accurately. Analyses suggested this mapped advantage was likely due to endpoint distinctiveness and the benefits of proximity errors during choice responding rather than a conceptual or relational advantage attributable to the common or ordinal mapping of the two dimensions. This potential difficulty in mapping relations across dimensions may limit the pigeons' capacity for more advanced types of analogical reasoning. This article is part of a Special Issue entitled: Tribute to Tom Zentall.

Auditory same/different concept learning and generalization in black-capped chickadees (Poecile atricapillus)

Abstract concept learning was thought to be uniquely human, but has since been observed in many other species. Discriminating same from different is one abstract relation that has been studied frequently. In the current experiment, using operant conditioning, we tested whether black-capped chickadees (Poecile atricapillus) could discriminate sets of auditory stimuli based on whether all the sounds within a sequence were the same or different from one another. The chickadees were successful at solving this same/different relational task, and transferred their learning to same/different sequences involving novel combinations of training notes and novel notes within the range of pitches experienced during training. The chickadees showed limited transfer to pitches that was not used in training, suggesting that the processing of absolute pitch may constrain their relational performance. Our results indicate, for the first time, that black-capped chickadees readily form relational auditory same and different categories, adding to the list of perceptual, behavioural, and cognitive abilities that make this species an important comparative model for human language and cognition.

Effects of stimulus size and spatial organization on pigeons' conditional same-different discrimination

In two experiments, we explored the effects of varying the size and the spatial organization of the stimuli in multi-item arrays on pigeons' same-different discrimination behavior. The birds had previously learned to discriminate a simultaneously presented array of 16 identical (Same) visual items from an array of 16 nonidentical (Different) visual items, when the correct choice was conditional on the presence of another cue: the color of the background (Castro et al., in press). In Experiment 1, we trained pigeons with 7-item arrays and then tested them with arrays containing the same item, but in a variety of sizes. In Experiment 2, we tested the birds with the items grouped in novel locations: the top, the bottom, the left, or the right portions of the display area, which generated different vertical and horizontal alignments. Accuracy scores revealed virtually perfect stimulus generalization across various item sizes and spatial organizations. Reaction times revealed that the birds perceived different sizes of a single icon as the same stimulus (Experiment 1) and that the birds processed vertical arrangements faster than horizontal arrangements (Experiment 2). These results suggest that the pigeons noticed both physical and spatial changes in the stimuli (as shown by their reaction times), but that these changes did not disrupt the birds' discriminating the sameness or differentness of the multi-item arrays (as shown by their accuracy scores).

Same/different discrimination learning with trial-unique stimuli

A long-standing issue in same/different discrimination learning concerns the possible role of individual stimulus memory through repeated presentation. The aim of eliminating any effect of repetition prompted us to devise a new method for generating trial-unique stimuli. These stimuli were arrays of 16 mosaics, each containing 16 cells, which could be filled with 16 possible luminance levels. In Experiment 1, we successfully trained 4 pigeons with these trial-unique stimuli in a two-alternative forced choice same/different discrimination task to 80% correct-choice performance. We later conducted two tests that explored the nature of this discrimination and suggested that pigeons compared the mosaics in the arrays on the basis of their spatial configurations, not on the basis of lower level perceptual properties. In Experiment 2, college students responded similarly to the same sequence of training and testing. Our results suggest that pigeons and people may use similar mechanisms in relational discrimination learning.

Learning and transfer of relational matching-to-sample by pigeons

We trained pigeons on a relational matching-to-sample task to see whether a nonprimate species can discriminate higher-order "relations between relations". W e required the birds t o relationally match arrays of 1 6 itemsthat were chosen from five nonoverlapping sets of 20 colored computer icons. On each trial, randomly selected icons from one set were placed into a 4 x 4 grid to form a sample; on same trials, all 16 icons were identical to each other, whereas on different trials, all 16 icons were different from each other. After 10-20 pecks, 16-item same and different testing arrays were presented that were created from a nentirely different icon set. Because noicons were common to the sample and testing arrays, discriminating higher-order relations was required for success on the tests. As have primates in similar tasks, pigeons successfully learned and transferred this relational discrimination, suggesting that both birds and mammals possess t he cognitive antecedents of analogicalreasoning.

Mental representation of symbols as revealed by vocabulary errors in two bonobos (Pan paniscus)

Error analysis has been used in humans to detect implicit representations and categories in language use. The present study utilizes the same technique to report on mental representations and categories in symbol use from two bonobos (Pan paniscus). These bonobos have been shown in published reports to comprehend English at the level of a two-and-a-half year old child and to use a keyboard with over 200 visuographic symbols (lexigrams). In this study, vocabulary test errors from over 10 years of data revealed auditory, visual, and spatio-temporal generalizations (errors were more likely items that looked like sounded like, or were frequently associated with the sample item in space or in time), as well as hierarchical and conceptual categorizations. These error data, like those of humans, are a result of spontaneous responding rather than specific training and do not solely depend upon the sample mode (e.g. auditory similarity errors are not universally more frequent with an English sample, nor were visual similarity errors universally more frequent with a photograph sample). However, unlike humans, these bonobos do not make errors based on syntactical confusions (e.g. confusing semantically unrelated nouns), suggesting that they may not separate syntactical and semantic information. These data suggest that apes spontaneously create a complex, hierarchical, web of representations when exposed to a symbol system.

Effects of number of items and visual display variability onsame-different discrimination behavior

We explored college students' discrimination of complex visual stimuli that involvedmultiple-item displays. The items in each of the displays could be all the same, all different, or diverse mixtures of some same and some different items. The participants had to learn which of two arbitrary responses was correct for each of the displays without being told about the sameness or differentness of the stimuli. We observed a general improvement in discrimination performance--a rise in choice accuracy and a fall in reaction time-as the number of icons in the display was increased, even when the participants had been trained from the outset with displays containing different numbers of items and when smaller numbers of items were not randomly distributed but grouped in the center of the display. The participants' discrimination behavior also depended on the mixture of same and different items in the displays. Striking individual differences in the participants' discrimination behavior disclosed that people sometimes respond as do pigeons and baboons trained with a similar task. This and previous related research suggest that variability discrimination may lie at the root of same-different categorization behavior.

Mechanisms of same/different concept learning in primates and avians

Mechanisms of same/different concept learning by rhesus monkeys, capuchin monkeys, and pigeons were studied in terms of how these species learned the task (e.g., item-specific learning versus relational learning) and how rapidly they learned the abstract concept, as the training set size was doubled. They had similar displays, training stimuli, test stimuli, and contingencies. The monkey species learned the abstract concept at similar rates and more rapidly than pigeons, thus showing a quantitative difference across species. All species eventually showed full concept learning (novel-stimulus transfer equivalent to baseline: 128-item set size for monkeys; 256-item set for pigeons), thus showing a qualitative similarity across species. Issues of stimulus regularity/symmetry, generalization from item pairs, and familiarity processing were not considered to be major factors in the final performances, converging on the conclusion that these species were increasingly controlled by the sample-test relationship (i.e., relational processing) leading to full abstract-concept learning.

Transposition in pigeons: reassessing Spence (1937) with multiple discrimination training

We studied transposition in pigeons, using multiple-pair discrimination training. Four birds discriminated two pairs of circles: 1+ 2- and 5+ 6- or 1 - 2 + and 5- 6 + (digits denote circle diameters and plus and minus signs denote reward and nonreward, respectively). Four other birds discriminated four pairs of circles: 1+ 2-, 1+ 3-, 4+ 6-, and 5+ 6- or 1- 2 +, 1- 3 +, 4- 6+, and 5- 6+. Finally, 4 birds discriminated only one pair of circles: 1+ 2-, 1- 2+, 5+ 6-, or 5- 6+. Testing included five new pairs--1/5, 2/3,2/6,3/4, and 4/5--that distinguished absolute from relational accounts of transposition. The pigeons' relational responding rose from one- to two- to four-pair training. The similarity of the testing stimuli to one another also affected relational responding: Transposition increased with highly dissimilar stimuli. Neither Spence's (1937) theory nor existing relational accounts could predict the obtained pattern of relational responding.

Time-course of control by specific stimulus features and relational cues during same-different discrimination training

We trained 7 pigeons to discriminate visual displays of 16 same items from displays of 16 different items. The specific stimulus features of the items and the relations among the items could serve as discriminative stimuli. Unlike in most studies of same-different discrimination behavior, we gave a small number of probe tests during each session of acquisition to measure the time-course of control by the learning of specific stimulus features and relational cues. Both the specific stimulus features and relational cues exerted reliable stimulus control, with the specific stimulus features exerting more control during the final three fourths of same-different learning. These findings replicate research suggesting that pigeons encode both the specific stimulus features and relational cues, and for the first time document the time-course of control by each kind of cue.

The pigeon's discrimination of visual entropy: a logarithmic function

We taught 8 pigeons to discriminate 16-icon arrays that differed in their visual variability or "entropy" to see whether the relationship between entropy and discriminative behavior is linear (in which equivalent differences in entropy should produce equivalent changes in behavior) or logarithmic (in which higher entropy values should be less discriminable from one another than lower entropy values). Pigeons received a go/no-go task in which the lower entropy arrays were reinforced for one group and the higher entropy arrays were reinforced for a second group. The superior discrimination of the second group was predicted by a theoretical analysis in which excitatory and inhibitory stimulus generalization gradients fall along a logarithmic, but not a linear scale. Reanalysis of previously published data also yielded results consistent with a logarithmic relationship between entropy and discriminative behavior.