Concept formation based on the relations among values of multi-component figures

A complex discrimination procedure was used to test class formation with multi-component figures in college students. First, selections of a red-+45 degrees -oriented rectangle (A1B1) instead of a red -45 degrees -oriented rectangle (A1B2) and of a green -45 degrees -oriented rectangle (A2B2) instead of a green-+45 degrees -oriented rectangle (A2B1), were reinforced. Second, selections of a +45 degrees -obtuse-angle-white hexagon (B1C1), instead of a +45 degrees -two-acute-angle hexagon (B1C2) and of a -45 degrees -two-acute-angle hexagon (B2C2), instead of a -45 degrees -obtuse-angle hexagon (B2C1), were reinforced. Subsequent tests with figures A1B1 and A2B1, A2B2 and A1B2, B1C1 and B2C1, and B2C2 and B1C2 demonstrated selection of the same figures as in training. Other tests with a novel figure compounded by color A1 and form C1 and a novel figure compounded by color A1 and form C2, or a novel figure compounded by A2 and C1 and a novel figure compounded by A2 and C2 demonstrated consistent selection of figures A1C1 and A2C2. A final test with figures A1C1 and A2C1, or with figures A2C2 and A1C2 also showed consistent selections of the same figures -A1C1 and A2C2. The resulting classification among selected and non-selected figures cannot be based on particular properties of the selected figures, or in a combination of a particular set of those properties, because both selected and non-selected figures were compounded by identical values. Instead, the participants selected the figures with two values of the same set (A1,B1, or C1, or A2,B2 or C2). Thus, these results demonstrated categorization based on the relations among the values of the figures. Therefore, these results have important implications for the study of concept formation.     

Stimulus class formation and functional equivalence in moderately retarded individuals' conditional discrimination

Three mentally retarded humans first acquired a simple discrimination: Simultaneously displayed visual stimuli A1 and A2 functioned as S+ and S-, respectively. The subjects also acquired a conditional discrimination, learning to select visual stimuli B1 and B2 conditionally upon A1 and A2, respectively. Then, B1 and B2 were displayed without A1 or A2. Subjects selected B1, an emergent discrimination that showed that B1 and B2 had become functionally equivalent to A1 and A2, respectively. Two subjects next learned to select C1 and C2 conditionally upon B1 and B2, respectively. They also learned to select B1 and B2 conditionally upon D1 and D2, respectively. Subsequent simple discrimination probe trials displayed (a) C1 and C2 and (b) D1 and D2. On the former, the subjects nearly always selected C1. On the latter, they initially selected D1 and D2 about equally often. Thus, the emergence of simple discrimination appears to depend on a specific experimental history.     

How robust are neural network models of stimulus generalization?

Artificial feed-forward neural networks are commonly used as a tool for modelling stimulus selection and animal signalling. A key finding of stimulus selection research has been generalization: if a given behaviour has been established to one stimulus, perceptually similar novel stimuli are likely to induce a similar response. Stimulus generalization, in feed-forward neural networks, automatically arises as a property of the network. This network property raises understandable concern regarding the sensitivity of the network to variation in its internal parameter values used in relation to its structure and to its training process. Researchers must have confidence that the predictions of their model follow from the underlying biology that they deliberately incorporated in the model, and not from often arbitrary choices about model implementation. We study how network training and parameter perturbations influence the qualitative and quantitative behaviour of a simple but general network. Specifically, for models of stimulus control we study the effect that parameter variation has on the shape of the generalization curves produced by the network. We show that certain network and training conditions produce undesirable artifacts that need to be avoided (or at least understood) when modelling stimulus selection.     

Judgments of shape similarity in the barbary dove (Streptopelia risoria)

Sixteen experimentally naive adult Barbary doves (S. risoria) learned a successive discrimination between a simple and a complex shape and were then tested for transfer of training in a generalization test with seven unfamiliar shapes. It was found that the degree of equivalence between shapes was correlated with two physical measures of shape based on contour length and on the number of sides of the shape. These results are in general agreement with other comparative data including those for humans. Results of the generalization test showed an ‘asymmetrical’ response in that judgments of shape similarity were not elicited with equal frequency in all subjects. This was interpreted as evidence for an internal standard in that subjects compared test stimuli to the positive training shape.     

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).     

The role of temporal generalization in a temporal discrimination task

Two experiments trained rats to discriminate two or three stimulus durations using a temporal discrimination task. A standard peak shift effect was observed when training was administered with short versus long signals in Experiment 1. Both discrimination accuracy scores and the generalization gradients revealed that shorter intervals were discriminated more accurately, which may be due to the scalar property of timing. In Experiment 2, three signals (short, medium, and long) were associated with three different responses, or two of the intervals were associated with one response (short and long or short and medium) and the other interval with a different response. Here, the short/medium versus long discrimination was learned most readily of the three tasks. The results of both experiments indicated a strong contribution of learning of individual durations combined with scalar generalization gradients, but Experiment 2 indicated that categorical encoding of durations may have also been operating.     

Effects of noise on the performance of rats in an operant discrimination task

Two experiments examined the effect of the presentation of an irregular, moderate intensity auditory stimulus ('noise') on the performance of rats in an operant discrimination task. In Experiment 1, rats first learned to press a lever in the presence of a visual stimulus but not in its absence. Discrimination performance was impaired during subsequent exposure to noise. In Experiment 2, different groups of rats learned the discrimination task under a noise or a no-noise condition. Thereafter, all rats were tested under each noise condition. Discrimination performance was best when the noise condition at test was identical to the noise condition at training. These results were discussed in the framework of arousal, distraction, generalization decrement, and contextual occasion setting. They point to the necessity of using a 2x2 factorial design in human and animal research on noise effects, with noise condition at training (noise present or absent) and noise condition at test (noise present or absent) as factors.     

Integrating memories in the human brain: hippocampal-midbrain encoding of overlapping events

Decisions are often guided by generalizing from past experiences. Fundamental questions remain regarding the cognitive and neural mechanisms by which generalization takes place. Prior data suggest that generalization may stem from inference-based processes at the time of generalization. By contrast, generalization may emerge from mnemonic processes occurring while premise events are encoded. Here, participants engaged in a two-phase learning and generalization task, wherein they learned a series of overlapping associations and subsequently generalized what they learned to novel stimulus combinations. Functional MRI revealed that successful generalization was associated with coupled changes in learning-phase activity in the hippocampus and midbrain (ventral tegmental area/substantia nigra). These findings provide evidence for generalization based on integrative encoding, whereby overlapping past events are integrated into a linked mnemonic representation. Hippocampal-midbrain interactions support the dynamic integration of experiences, providing a powerful mechanism for building a rich associative history that extends beyond individual events.     

An image-matching mechanism describes a generalization task in honeybees

Currently two hypotheses exist as to how an insect's CNS stores and compares learned visual images: storage as a photograph-like eidetic image (template-hypothesis), and storage as a set of features extracted by the visual system (parameter-hypothesis). An important criterion for deciding between these two hypotheses is the generalization of a learned pattern to other patterns. We therefore trained bees to patterns presented on a vertical screen and to discriminate between either a black triangle and a ring-pattern (training A, B), or a black disc and an outline triangle (training C), and then tested for a possible generalization of the learned patterns to a variety of other patterns. Several results obtained that way (e.g. the transfer test, column 2 in Fig. 1a, or Fig. 3) seem to indicate that the bees indeed do generalize a learned pattern to other patterns, and hence appear to support the parameter-hypothesis. Quite unexpectedly, however, it turned out that the same results can also be explained by a template matching mechanism. In simulating an image matching strategy we used a similarity function that had been successfully applied to pattern discrimination data obtained in tetheredly flying fruitflies, Drosophila. This similarity function assumes that the insect evaluates the ratio between the overlap of an actual retinal image with the stored template and the total area of the actual retinal image (see Fig. 5). It is able to describe the bulk of our data. Interestingly, the training situation appears to have an influence on the weight attributed by the bee to different parts of the template. Such a situation-dependent weighting might indicate a first stage of parameter extraction.Abbreviations TR + positive training shape - TR negative training shape     

Generalization of motor learning depends on the history of prior action

Generalization of motor learning refers to our ability to apply what has been learned in one context to other contexts. When generalization is beneficial, it is termed transfer, and when it is detrimental, it is termed interference. Insight into the mechanism of generalization may be acquired from understanding why training transfers in some contexts but not others. However, identifying relevant contextual cues has proven surprisingly difficult, perhaps because the search has mainly been for cues that are explicit. We hypothesized instead that a relevant contextual cue is an implicit memory of action with a particular body part. To test this hypothesis we considered a task in which participants learned to control motion of a cursor under visuomotor rotation in two contexts: by moving their hand through motion of their shoulder and elbow, or through motion of their wrist. Use of these contextual cues led to three observations: First, in naive participants, learning in the wrist context was much faster than in the arm context. Second, generalization was asymmetric so that arm training benefited subsequent wrist training, but not vice versa. Third, in people who had prior wrist training, generalization from the arm to the wrist was blocked. That is, prior wrist training appeared to prevent both the interference and transfer that subsequent arm training should have caused. To explain the data, we posited that the learner collected statistics of contextual history: all upper arm movements also move the hand, but occasionally we move our hands without moving the upper arm. In a Bayesian framework, history of limb segment use strongly affects parameter uncertainty, which is a measure of the covariance of the contextual cues. This simple Bayesian prior dictated a generalization pattern that largely reproduced all three findings. For motor learning, generalization depends on context, which is determined by the statistics of how we have previously used the various parts of our limbs.     

Effects of conditioning history on selective stimulus control by elements of compound discriminative stimuli

Effects of prior discrimination training on stimulus control by color and shape dimensions of compound stimuli were studied with college students. In Phase 1, single-stimulus discrimination training was conducted for two values of color and shape. Phase 2 discrimination training employed two 2-dimensional compound stimuli composed of the color and shape stimuli trained in Phase 1. For conflict-compound stimuli, the stimulus-response-consequence contingency was altered between phases for one stimulus dimension (target dimension), but not for the other, non-target, dimension. Level of congruence (100%, 25%, and 0%) of the contingency for the target dimension between phases was manipulated across groups. When each stimulus value was tested in Phase 3, level of Phase-2-consistent responding to the target dimension varied with level of Phase-1-to-Phase-2 congruence. In Experiment 2, training history for the non-target dimension was altered across three conditions: (a) Correlated with reinforcement, as in Experiment 1, (b) No-Training, or (c) Not-Correlated. Phase-2-consistent responding to the target cue in Phase 3 was lower under the latter conditions than under the Correlated condition, indicating that the non-target dimension modulated control by the target dimension, consistent with stimulus competition. The data suggest elemental, rather than configural processing of the compound stimuli during Phase 2.     

Orientation discrimination and categorization of photographs of natural objects by pigeons

In Experiment 1, pigeons trained to discriminate rightside-up and upside-down orientations of slides of natural scenes with humans successfully transferred to new slides of the same kind. Experiment 2 revealed that both the orientations of the human figures and of the background scenes controlled the discrimination. When they were oppositely oriented, the background orientation cue was dominant. In Experiment 3 slides showing objects on a white background were presented either rightside up or upside down, with each slide presented in one orientation only. One group of pigeons learned to classify the slides according to their orientations. The other group learned to classify the slides according to arbitrary groupings. When the slides were shown rotated by 180 degrees, the latter group continued to discriminate the individual slides (i.e., the pigeons showed orientation invariance). The former group classified the rotated slides according to their orientations (i.e., orientation discrimination). In Experiment 4, pigeons learned the orientation discrimination with separate sets of human and bird figures. Partial reversal training in one object class transferred to the rest of stimuli in this object class but did not to the other object class. These results suggest that pigeons can learn to discriminate photographs on the basis of orientation but that orientation-based equivalence relationship is not formed between object classes.     

Lack of evidence for inhibitory processes in over-selectivity

Stimulus over-selectivity can be defined as control over behavior being exerted by one aspect of the environment at the expense of other equally salient aspects of the environment, and is a common problem for discrimination learning under conditions of cognitive strain, and in intellectual disorders. Non-clinical participants exposed to a concurrent task load were trained and tested on a two-component trial-and-error discrimination task to investigate whether inhibition plays a role in producing under-selectivity by using both summation and retardation tests. Experiment 1 found evidence for the over-selectivity effect, and replicated the finding that revaluation of a previously over-selected stimulus allows emergence of control by a previous under-selected stimulus, despite the latter stimulus receiving no direct conditioning. The under-selected cue was not found to gain any conditioned inhibitory status, as reflected by summation (Experiment 2), and retardation (Experiment 3), tests. The results extend the literature explaining over-selectivity as a post-acquisition failure.     

Enhancing biomedical text summarization using semantic relation extraction

Automatic text summarization for a biomedical concept can help researchers to get the key points of a certain topic from large amount of biomedical literature efficiently. In this paper, we present a method for generating text summary for a given biomedical concept, e.g., H1N1 disease, from multiple documents based on semantic relation extraction. Our approach includes three stages: 1) We extract semantic relations in each sentence using the semantic knowledge representation tool SemRep. 2) We develop a relation-level retrieval method to select the relations most relevant to each query concept and visualize them in a graphic representation. 3) For relations in the relevant set, we extract informative sentences that can interpret them from the document collection to generate text summary using an information retrieval based method. Our major focus in this work is to investigate the contribution of semantic relation extraction to the task of biomedical text summarization. The experimental results on summarization for a set of diseases show that the introduction of semantic knowledge improves the performance and our results are better than the MEAD system, a well-known tool for text summarization.     

Use of current-voltage diagrams in locating peak energy barriers in cell membranes

Summary The current-voltage relations obtained by integrating the Nernst-Planck equations for a variety of energy profiles are obtained. A simple and approximate method for comparing these relations is described. The method is based on using a linearized transform of current-voltage relations for an Eyring single barrier model. A parameter, , related to the location of the single barrier in the Eyring model, and to the shape of the barrier in other models, is readily obtained from the slopes of the linearized relations. It is then a simple matter to determine whether a given current-voltage relation allows discrimination between any particular energy profiles. The results show that the equivalent Eyring model does not always place the peak energy barrier in the same position as other models and that quite large errors in the assignment of position may be made if such a model is used. The results are also used to test the ability of some experimental current-voltage diagrams to discriminate between various energy profiles.     

Qualitative discrimination of sweet stimuli: behavioural study on rats

Rats conditioned to avoid drinking a sucrose solution or a glucosesolution were tested for generalization of the learned aversionto three lower concentrations of these sweeteners. The behaviouralresponses indicated two types of discrimination. The first one,after two seconds, is of a qualitative nature; the second one,after five seconds, is of a quantitative nature showing an aversiveresponse depending on the concentration. Cross-aversion to tengustatory stimuli showed on the basis of the qualitative discriminationthat the rats discriminate between the different sweetenersin a similar way as man does.     

Recognition of visual images by separate hemispheres in conditions of masked blocking of interhemispheric transmission

The subjects learned to recognize three figures presented in the left visual hemifield and three figures presented in the right visual hemifield. During presentation of a stimulus, the contralateral hemifield was overlapped by a mask. After the training, recognition of all six figures presented in the right and left visual hemifields, was compared. Each hemisphere recognizes figures which were learned in the corresponding visual hemifield, but the recognition of figures learned in the opposite visual hemifield was poor. Thus, the ability of the hemispheres to act separately in recognizing different sets of visual images, was established.     

Matching unrelated stimuli with same discriminative functions: training order effects

Previous research has shown that after training simple discriminations (A1+/A2−, B1+/B2−), bringing these tasks under conditional control (J1–A1, J2–A2) leads to transfer of discriminative control (J1+/J2−) and to generalized matching on the basis of same discriminative functions (e.g. J1–B1, J2–B2). The same occurs when conditional discriminations are trained (D1–E1, D2–E2; F1–G1, F2–G2). When the subjects are then trained to demonstrate correct relations (D1–E1, D2–E2) when given X1 and to demonstrate incorrect relations when given X2 (XD–E), transfer of discriminative control (X1+/X2−) and generalized matching on the basis of same discriminative functions emerges (e.g. X1F1–G1, X2F1–G2). The present study investigated if these performances are dependent on the training and/or testing order. In Experiment 1, the lower-order contingency tasks were trained before the higher-order contingency tasks (A1+/A2−, B1+/B2− before J–A, and D–E, F–G before XD–E). Half the subjects received the J–B test before the more complex XF–G test (Condition A), while for the other subjects, this testing order was reversed (Condition B). Finally, all subjects received additional tests in which they were given the opportunity to demonstrate the discriminative properties of the J and X stimuli (J1+/J2−, X1+/X2−), and to match the A, J, and X stimuli with newly introduced stimuli of same discriminative properties (e.g. J1-POLITE, J2-RUDE). Experiment 2 was the same except that the training order was reversed (J–A before A1+/A2−, B1+/B2−, and XD–E before D–E, F–G). The results were affected by the training order but not by the testing order. Transfer of discriminative functions and generalized matching on the basis of same functions only occurred reliably when the lower-order contingency tasks were trained first. A stimulus-control account of the data is offered.     

The effect of stimulus discriminability on strategies for learning multiple temporal discriminations

The goal was to identify training conditions under which temporal intervals that are signaled by different stimuli are memorized (i.e., the temporal behavior is readily shown to be under stimulus control). Undergraduate students were trained on three signaled temporal discriminations using a peak procedure. The intervals were trained in either blocks of trials or with trials intermixed within the session, and then they were given a transfer test with intermixed trials. There were two levels of stimulus discriminability, defined by the similarity of the stimuli. Most participants memorized the intervals when the discriminations were intermixed within the session, or were easy, but not when the discriminations occurred in blocks and were difficult. In the transfer tests, those participants trained in the difficult discrimination that occurred in blocks of trials typically continued to perform as they did during the last-trained interval, regardless of the stimulus presented. These results are better explained by a memory retrieval than a memory storage account.     

Untersuchungen zum Symmetriekonzept bei Kohlmeisen (Parus major L.)*

The question was investigated whether an insectivorous bird when hungry can form the concept of bilateral symmetry which might enable it to detect resting prey proficiently. Eight great tits of both sexes were trained in four experiments with insects as reward to discriminate either between a pair of figures that differed only in terms of the presence of bilateral symmetry (symmetric—asymmetric), or simultaneously between pairs of figures that were both symmetric (symmetric—symmetric) or not (symmetric—asymmetric). Symmetry concept formation was demonstrated in some of the Ss by a) generalization tests and b) a better performance with symmetric-asymmetric pairs as compared to symmetric—symmetric pairs. While one S attended to symmetry regardless of the orientation of the symmetry axis, another probably did not.