Peck tracking: a method for localizing critical features within complex pictures for pigeons

The pigeon is a standard animal in comparative psychology and is frequently used to investigate visuocognitive functions. Nonetheless, the strategies that pigeons use to discriminate complex visual stimuli remain a difficult area of study. In search of a reliable method to identify features that control the discrimination behaviour, pecking location was tracked using touch screen technology in a people-absent/people-present discrimination task. The correct stimuli contained human figures anywhere on the picture, but the birds were not required to peck on that part. However, the stimuli were designed in a way that only the human figures contained distinguishing information. All pigeons focused their pecks on a subarea of the distinctive human figures, namely the heads. Removal of the heads significantly impaired performance, while removal of other distinctive parts did not. Thus, peck tracking reveals the location within a complex visual stimulus that controls discrimination behaviour, and might be a valuable tool to reveal the strategies pigeons apply in visual discrimination tasks.     

Learning efficient visual search for stimuli containing diagnostic spatial configurations and color-shape conjunctions

Visual search is often slow and difficult for complex stimuli such as feature conjunctions. Search efficiency, however, can improve with training. Search for stimuli that can be identified by the spatial configuration of two elements (e.g., the relative position of two colored shapes) improves dramatically within a few hundred trials of practice. Several recent imaging studies have identified neural correlates of this learning, but it remains unclear what stimulus properties participants learn to use to search efficiently. Influential models, such as reverse hierarchy theory, propose two major possibilities: learning to use information contained in low-level image statistics (e.g., single features at particular retinotopic locations) or in high-level characteristics (e.g., feature conjunctions) of the task-relevant stimuli. In a series of experiments, we tested these two hypotheses, which make different predictions about the effect of various stimulus manipulations after training. We find relatively small effects of manipulating low-level properties of the stimuli (e.g., changing their retinotopic location) and some conjunctive properties (e.g., color-position), whereas the effects of manipulating other conjunctive properties (e.g., color-shape) are larger. Overall, the findings suggest conjunction learning involving such stimuli might be an emergent phenomenon that reflects multiple different learning processes, each of which capitalizes on different types of information contained in the stimuli. We also show that both targets and distractors are learned, and that reversing learned target and distractor identities impairs performance. This suggests that participants do not merely learn to discriminate target and distractor stimuli, they also learn stimulus identity mappings that contribute to performance improvements.     

Symmetry and stimulus class formation in humans: Control by temporal location in a successive matching task

Symmetry refers to the observation that subjects will derive B‐A (e.g., in the presence of B, select A) after being trained on A‐B (e.g., in the presence of A, select B). Whereas symmetry is readily shown in humans, it has been difficult to demonstrate in nonhuman animals. This difficulty, at least in pigeons, may result from responding to specific stimulus properties that change when sample and comparison stimuli switch roles between training and testing. In three experiments with humans, we investigated to what extent human responding is influenced by the temporal location of stimuli using a successive matching‐to‐sample procedure. Our results indicate that temporal location does not spontaneously control responding in humans, although it does in pigeons. Therefore, the number of functional stimuli that humans respond to in this procedure may be half of the number of functional stimuli that the pigeons respond to. In a fourth experiment, we tested this assumption by doubling the number of functional stimuli controlling responding in human participants in an attempt to make the test more comparable to symmetry tests with pigeons. Here, we found that humans responded according to indirect class formation in the same manner as pigeons do. In sum, our results indicate that functional symmetry is readily observed in humans, even in cases where the temporal features of the stimuli prevent functional symmetry in pigeons. We argue that this difference in behavior between the two species does not necessarily reflect a difference in capacity to show functional symmetry between both species, but could also reflect a difference in the functional stimuli each species responds to.     

Category discrimination by pigeons using five polymorphous features

Eight pigeons were trained to discriminate between sets of color photographs of natural scenes. The scenes differed along five two-valued dimensions (site, weather, camera distance, camera orientation, and camera height), and all combinations of the feature values were used. One value of each dimension was designated as positive, and slides containing three or more positive feature values were members of the positive stimulus set. Thus, each feature had an equal, low, correlation with reinforcement, and all features had zero correlations with each other. Seven of the 8 pigeons learned this discrimination, and their responding came under the control of all five features. Within the positive and negative stimulus sets, response rates were higher to stimuli that contained more positive feature values. Once discrimination had been achieved, reversal training was given using a subset of the slides. In this subset, only a single feature was correlated with reinforcement. All pigeons learned this reversal successfully and generalized it to additional photographs with the same feature content. After reversal, the original reinforcement contingencies were reinstated, and training was continued using all the slides except those that had been used in reversal. Reversal generalized to these slides to some extent. Analysis of the response rates to individual slides showed that, compared with prereversal training, only the feature that had been subjected to reversal contingencies showed a reversed correlation with response rate. The remaining features showed the same correlation with response rate as they had before reversal training. Thus, reversal on some members of a category following category discrimination training led to generalization to stimuli within the category that were not involved in the reversal, but not to features that were not reversed. It is therefore inappropriate to refer to the pigeons as learning a concept.     

Categorical templates are more useful when features are consistent: Evidence from eye movements during search for societally important vehicles

Unlike in laboratory visual search tasks—wherein participants are typically presented with a pictorial representation of the item they are asked to seek out—in real-world searches, the observer rarely has veridical knowledge of the visual features that define their target. During categorical search, observers look for any instance of a categorically defined target (e.g., helping a family member look for their mobile phone). In these circumstances, people may not have information about noncritical features (e.g., the phone’s color), and must instead create a broad mental representation using the features that define (or are typical of) the category of objects they are seeking out (e.g., modern phones are typically rectangular and thin). In the current investigation (Experiment 1), using a categorical visual search task, we add to the body of evidence suggesting that categorical templates are effective enough to conduct efficient visual searches. When color information was available (Experiment 1a), attentional guidance, attention restriction, and object identification were enhanced when participants looked for categories with consistent features (e.g., ambulances) relative to categories with more variable features (e.g., sedans). When color information was removed (Experiment 1b), attention benefits disappeared, but object recognition was still better for feature-consistent target categories. In Experiment 2, we empirically validated the relative homogeneity of our societally important vehicle stimuli. Taken together, our results are in line with a category-consistent view of categorical target templates (Yu, Maxfield, & Zelinsky in, Psychological Science, 2016. doi: 10.1177/0956797616640237), and suggest that when features of a category are consistent and predictable, searchers can create mental representations that allow for the efficient guidance and restriction of attention as well as swift object identification.     

Reflexivity without identity matching training: A first demonstration

Until now, the equivalence property of reflexivity—matching physically identical stimuli to themselves after training on a set of arbitrary matching relations—has not been demonstrated in any animal, human or nonhuman. Previous reports of reflexivity have either implicitly or explicitly involved reinforced training on other identity matching relations. Here we demonstrate reflexivity without prior identity matching training. Pigeons received concurrent successive matching training on three arbitrary matching tasks: AB (hue–form), BC (form–hue), and AC (hue–hue with different hues in the A and C sets). Afterwards, pigeons were tested for BB (form–form) reflexivity. Consistent with the predictions of Urcuioli's ( 2008 ) theory, pigeons preferentially responded to B comparison stimuli that matched the preceding B sample stimuli in testing (i.e., BB reflexivity). A separate experiment showed that a slightly different set of arbitrary matching baseline relations yielded a theoretically predicted “anti‐reflexivity” (or emergent oddity) effect in two of five pigeons. Finally, training on just two arbitrary successive matching tasks (AB and BC) did not yield any differential BB responding in testing for five of eight pigeons, with two others showing reflexivity and one showing antireflexivity. These data complement previous findings of symmetry and transitivity (the two other properties of equivalence) in pigeons.     

Is a Pink Cow Still a Cow? Individual Differences in Toddlers' Vocabulary Knowledge and Lexical Representations

When a toddler knows a word, what does she actually know? Many categories have multiple relevant properties; for example, shape and color are relevant to membership in the category banana . How do toddlers prioritize these properties when recognizing familiar words, and are there systematic differences among children? In this study, toddlers viewed pairs of objects associated with prototypical colors. On some trials, objects were typically colored (e.g., Holstein cow and pink pig); on other trials, colors were switched (e.g., pink cow and Holstein‐patterned pig). On each trial, toddlers were directed to find a target object. Overall, recognition was disrupted when colors were switched, as measured by eye movements. Moreover, individual differences in vocabularies predicted recognition differences: Toddlers who say fewer shape‐based words were more disrupted by color switches. “Knowing” a word may not mean the same thing for all toddlers; different toddlers prioritize different facets of familiar objects in their lexical representations.     

Redundancy gain for semantic features

In a go/no-go experiment, semantic redundancy gain was assessed for responses to single written words. Specifically, we asked participants to respond only to words whose meaning matched at least one semantic target feature—that is, the target category (e.g., animal), the target color (e.g., gray), or both. On redundant-target trials, the word (e.g., elephant) matched both semantic target features (i.e., gray and animal). On single-target trials, the word (e.g., beaver) matched one target feature (i.e., animal) and a nontarget feature (i.e., brown). We observed shorter reaction times in the redundant-target condition than in the faster single-target condition. Hence, the present study provides the first evidence that redundancy gain is not limited to responses to redundant proximal stimulus features but can also be observed for responses to semantic feature information.     

The Effect of Evidential Impact on Perceptual Probabilistic Judgments

In a series of three behavioral experiments, we found a systematic distortion of probability judgments concerning elementary visual stimuli. Participants were briefly shown a set of figures that had two features (e.g., a geometric shape and a color) with two possible values each (e.g., triangle or circle and black or white). A figure was then drawn, and participants were informed about the value of one of its features (e.g., that the figure was a “circle”) and had to predict the value of the other feature (e.g., whether the figure was “black” or “white”). We repeated this procedure for various sets of figures and, by varying the statistical association between features in the sets, we manipulated the probability of a feature given the evidence of another (e.g., the posterior probability of hypothesis “black” given the evidence “circle”) as well as the support provided by a feature to another (e.g., the impact, or confirmation, of evidence “circle” on the hypothesis “black”). Results indicated that participants’ judgments were deeply affected by impact, although they only should have depended on the probability distributions over the features, and that the dissociation between evidential impact and posterior probability increased the number of errors. The implications of these findings for lower and higher level cognitive models are discussed.     

On the Interaction of Prior Knowledge and Stimulus Structure in Category Learning

Contemporary theories of categorization propose that concepts are coherent in virtue of being embedded in a network of theories about the world. Those theories function to pick out some of the many possible features of a set of objects as most salient for purposes of classification, a process that is complex and still poorly understood (Murphy & Medin, 1985). Part of what makes this account incomplete is a lack of information as to (1) what makes a feature salient on a given occasion and (2) how feature salience interacts with category structure to determine the course of learning. We report on the results of three studies of category learning using complex schematic drawings to show that (1) the contrast set defined by one's initial encounters with category exemplars can be a source of individual differences in feature salience assignments; (2) such effects are short-lived in the face of clear evidence about actual feature diagnosticity; and (3) more robust prior hypotheses interact with category structure to either enhance learning or impede it. The enhancement occurs when the hypothesis emphasizes category-relevant features, even if the hypothesis is in fact incorrect. A hypothesis that assigns high salience to irrelevant features impedes learning. Learning does occur as feedback concerning category structure leads to enhanced salience for relevant features. Salience of irrelevant features remains high, however, suggesting that such learning as occurs involves augmentation and not total revision of the (incorrect) prior hypothesis.     

Contextual cuing by global features

In visual search tasks, attention can be guided to a target item--appearing amidst distractors--on the basis of simple features (e.g., finding the red letter among green). Chun and Jiang's (1998) contextual cuing effect shows that reaction times (RTs) are also speeded if the spatial configuration of items in a scene is repeated over time. In the present studies, we ask whether global properties of the scene can speed search (e.g., if the display is mostly red, then the target is at location X). In Experiment 1A, the overall background color of the display predicted the target location, and the predictive color could appear 0, 400, or 800 msec in advance of the search array. Mean RTs were faster in predictive than in nonpredictive conditions. However, there was little improvement in search slopes. The global color cue did not improve search efficiency. Experiments 1B-1F replicated this effect using different predictive properties (e.g., background orientation-texture and stimulus color). The results showed a strong RT effect of predictive background, but (at best) only a weak improvement in search efficiency. A strong improvement in efficiency was found, however, when the informative background was presented 1,500 msec prior to the onset of the search stimuli and when observers were given explicit instructions to use the cue (Experiment 2).     

Does Conventional Implicit Association Test of Shyness Measure “Self‐Shyness” or “Others‐Shyness”?

The Implicit Association Test (IAT) measures implicit associations between attitude targets and attributes. Its structure and procedure facilitate investigation of the strength of associations between one target and attributes relative to that between the other target and the same attributes when two targets are contradictory (e.g., black/white and comfortable/uncomfortable). This structure can cause conceptual complexity about what the IAT measures, particularly when a counter category is not needed. Thus, using the Single‐Target Implicit Association Test (ST‐IAT), which allowed only one target category for pairing with attributes, this paper delineated the association measured in the conventional IAT for shyness: “self‐shy” or “others‐shy.” Seventy‐seven Japanese university students completed the self‐report shyness scale, the conventional IAT, and two ST‐IATs (i.e., self/others as target). Results showed that implicit shyness produced in the conventional IAT significantly and positively correlated with that in the self‐targeted ST‐IAT. Moreover, implicit shyness in the conventional IAT was significantly accounted for by those produced by the ST‐IAT with self as target and those calculated in the ST‐IAT with others as target in opposite directions.     

Generalization of the disruptive effects of alternative stimuli when combined with target stimuli in extinction

Differential‐reinforcement treatments reduce target problem behavior in the short term but at the expense of making it more persistent long term. Basic and translational research based on behavioral momentum theory suggests that combining features of stimuli governing an alternative response with the stimuli governing target responding could make target responding less persistent. However, changes to the alternative stimulus context when combining alternative and target stimuli could diminish the effectiveness of the alternative stimulus in reducing target responding. In an animal model with pigeons, the present study reinforced responding in the presence of target and alternative stimuli. When combining the alternative and target stimuli during extinction, we altered the alternative stimulus through changes in line orientation. We found that (1) combining alternative and target stimuli in extinction more effectively decreased target responding than presenting the target stimulus on its own; (2) combining these stimuli was more effective in decreasing target responding trained with lower reinforcement rates; and (3) changing the alternative stimulus reduced its effectiveness when it was combined with the target stimulus. Therefore, changing alternative stimuli (e.g., therapist, clinical setting) during behavioral treatments that combine alternative and target stimuli could reduce the effectiveness of those treatments in disrupting problem behavior.     

Differential brain responses when applying criterion attribute versus family resemblance rule learning

Subsystems of category learning have been identified on the basis of general domains of content (e.g., tools, faces). The present study examined categories from the standpoint of internal structure and determined brain topography associated with expressing two fundamentally different category rule structures (criterion attribute, CA, and family resemblance, FR). CA category learning involves processing stimuli by isolated features and classifying by properties held by all members. FR learning involves processing stimuli by integral wholes and classifying on overall similarity among members without sharing identical features. fMRI BOLD response to CA and FR categorization was measured with pseudowords as stimuli. Category knowledge for both tasks was mastered prior to brain imaging. Areas of activation emerged unique to the structure of each category and followed from the nature of the rule abstraction procedure. CA categorization was implemented by strong target monitoring and expectation (medial parietal), rule maintenance in working memory, feature selection processes (inferior frontal), and a sensitivity to high frequency components of the stimulus such as isolated features (anterior temporal). FR categorization, consistent with its multi-featural nature, involved word-level processing (left extrastriate) that evoked articulatory rehearsal (medial cerebellar). The data suggest category structure is an important determinant of brain response during categorization. For instance, anterior temporal structures may help attune visual processing systems to high frequency components to support the learning of criterial, highly predictive rules.     

Escape from serial stimuli leading to food

If the functional relations governing the strength of a conditioned reinforcer correspond to those obtained with other Pavlovian procedures (e.g., Kaplan, 1984), the termination of stimuli appearing early in the interval between successive food deliveries should be reinforcing. During initial training we presented four key colors, followed by food, in a recurrent sequence to each of 6 pigeons. This established a baseline level of autoshaped pecking. In later sessions, we terminated each of these colors or only the first color for a brief period following each peck, replacing the original color with a standard substitute to avoid darkening the key. Pecking decreased in the presence of the last color in the sequence but increased in the presence of the first. In accord with contemporary models of Pavlovian conditioning, these and other data suggest that the behavioral effects of stimuli in a chain may be better understood in terms of what each stimulus predicts, as measured by relative time to the terminal reinforcer, than in the exclusively positive terms of the traditional formulation (Skinner, 1938). The same model may also account for the initial pause under fixed-interval and fixed-ratio schedules of reinforcement.     

The contribution of category relationships to similarity ratings.

Two experiments produced evidence that category relationships contribute to similarity ratings. In Experiment 1, participants gave similarity ratings with respect to a semantic category (VEGETABLE) and a set of exemplars, some of which were members of the category (e.g., BROCCOLI) and some of which were not (e.g., BANANA). A regression analysis was used to predict the similarity ratings in terms of numbers of common and distinctive features, as reported by other participants. Perceived similarity was greater for examplars that were members of the category, independently of feature overlap. Experiment 2 examined similarity ratings with respect to pairs of exemplars. In some cases, both exemplars were members of the same category (e.g., BROCCOLI/CUCUMBER). In other cases, one exemplar was a member of the category and the other was not (e.g., BROCCOLI/BANANA). A regression analysis was used to predict the similarity ratings in terms of numbers of common and distinctive features. Perceived similarity was greater when both exemplars were members of the same category, independently of feature overlap.     

Lateralized cognition: asymmetrical and complementary strategies of pigeons during discrimination of the "human concept"

This study was aimed at revealing which cognitive processes are lateralized in visual categorizations of "humans" by pigeons. To this end, pigeons were trained to categorize pictures of humans and then tested binocularly or monocularly (left or right eye) on the learned categorization and for transfer to novel exemplars (Experiment 1). Subsequent tests examined whether they relied on memorized features or on a conceptual strategy, using stimuli composed of new combinations of familiar and novel humans and backgrounds (Experiment 2), whether the hemispheres processed global or local information, using pictures with different levels of scrambling (Experiment 3), and whether they attended to configuration, using distorted human figures (Experiment 4). The results suggest that the left hemisphere employs a category strategy and concentrates on local features, while the right hemisphere uses an exemplar strategy and relies on configuration. These cognitive dichotomies of the cerebral hemispheres are largely shared by humans, suggesting that lateralized cognitive systems already defined the neural architecture of the common ancestor of birds and mammals.     

Reorientation in a two-dimensional environment: II. Do pigeons (Columba livia) encode the featural and geometric properties of a two-dimensional schematic of a room?

Pigeons (Columba livia) searched for a hidden target area in images showing a schematic rectangular environment. The absolute position of the goal varied across trials but was constant relative to distinctive featural cues and geometric properties of the environment. Pigeons learned to use both of these properties to locate the goal. Transformation tests showed that pigeons could use either the color or shape of the features, but performance was better with color cues present. Pigeons could also use a single featural cue at an incorrect corner to distinguish between the correct corner and the geometrically equivalent corner; this indicates that they did not simply use the feature at the correct corner as a beacon. Interestingly, pigeons that were trained with features spontaneously encoded geometry. The encoded geometric information withstood vertical translations but not orientation transformations.     

Number-of-features effects and semantic processing

Participants list many semantic features for some concrete nouns (e.g., lion) and fewer for others (e.g., lime; McRae, de Sa, & Seidenberg, 1997). Pexman, Lupker, and Hino (2002) reported faster lexical decision and naming responses for high number of features (NOF) words than for low-NOF words. In the present research, we investigated the impact of NOF on semantic processing. We observed NOF effects in a self-paced reading task when prior context was not congruent with the target word (Experiment 1) and in a semantic categorization task (concrete vs. abstract; Experiment 2A). When we narrowed our stimuli to include high- and low-NOF words from a single category (birds), we found substantial NOF effects that were modulated by the specificity of the categorization task (Experiments 3A, 3B, and 3C). We argue that these results provide support for distributed representation of word meaning.     

Categorical congruence facilitates multisensory associative learning

Learning about objects often requires making arbitrary associations among multisensory properties, such as the taste and appearance of a food or the face and voice of a person. However, the multisensory properties of individual objects usually are statistically constrained, such that some properties are more likely to co-occur than others, on the basis of their category. For example, male faces are more likely to co-occur with characteristically male voices than with female voices. Here, we report evidence that these natural multisensory statistics play a critical role in the learning of novel, arbitrary associative pairs. In Experiment 1, we found that learning of pairs consisting of human voices and gender-congruent faces was superior to learning of pairs consisting of human voices and gender-incongruent faces or of pairs consisting of human voices and pictures of inanimate objects (plants and rocks). In Experiment 2, we found that this “categorical congruency” advantage extended to nonhuman stimuli, as well—namely, to pairs of class-congruent animal pictures and vocalizations (e.g., dogs and barks) versus class-incongruent pairs (e.g., dogs and bird chirps). These findings suggest that associating multisensory properties that are statistically consistent with the various objects that we encounter in our daily lives is a privileged form of learning.