Abstract: In experiment 1, pigeons were trained to match temporal (2, 8, and 10 s of houselight) and location (feeder light, left key, center key illumination) samples to color comparison stimuli. Red choices were correct following the 2-s and feeder light, orange choices were correct following the 8-s and center key, and green choices were correct following the 10-s and left key. Samples that were harder to discriminate (8- vs 10-s, and left vs center key) were mapped onto comparisons that were easy to discriminate (orange vs green), while samples that were easier to discriminate (2- vs 8-s, and feeder light vs left key) were mapped onto comparisons that were hard to discriminate(red vs orange). The pattern of errors for temporal and location samples indicated that these samples were not represented by a common code even though they were associated with the same comparison stimuli. In experiment 2, the same pigeons were trained with visual samples in which samples that were hard to discriminate (triangle vs circle) were mapped onto comparisons that were easy to discriminate (orange vs green), while samples that were easy to discriminate(plus vs triangle) were mapped onto comparisons that were hard to discriminate (red vs orange). Following acquisition of the visual discrimination, the temporal samples were re-introduced and many-to-one training was continued. During delay testing, the pattern of errors for temporal and visual samples was equivalent and consistent with the hypothesis that visual samples were being coded in terms of the duration appropriate for the temporal sample with which it shared a common comparison response. Data from no-sample test sessions ruled out a simple response bias explanation of the data. The properties of common codes for temporal and nontemporal events can be somewhat flexible and more complicated than previously envisaged.

Abstract: Five hens, experienced in discrimination of two categories of multidimensional geometrical figures presented in fixed pairs in a simultaneous discrimination, were tested with familiar figures arranged as new pairs to assess the dependence of categorization performance on learned relational or configural cues. Test performance did not differ from training: relational or configural cues still influenced discrimination performance. It was suggested that – in accordance with exemplar theories – this influence depended on differences between pairs of probe exemplars that facilitate retrieval of learned category members. To test whether exemplar, feature or prototype theory was most suitable to explain categorization by chickens, the rates of pecking at exemplars were analysed using principal components analysis (PCA). The distribution of the exemplars' component loads on the single component obtained was examined in the light of the conditions dictated by the three types of theories on how representative category exemplars should be. The least constraining theory, i.e. the exemplar theory, was most suitable. Defining factors of classificatory behaviour are discussed with a special emphasis on the characteristics of category-defining stimulus attributes.

Abstract: Day-old chicks were exposed to point-light animation sequences depicting either a walking hen or a rotating cylinder. On a subsequent free-choice test (experiment 1) the chicks approached the novel stimulus, irrespective of this being the hen or the cylinder. In order to obtain equivalent local motion vectors, in experiments 2 and 3 newly hatched chicks were exposed either to a point-light animation sequence depicting a walking hen, or to a positionally scrambled walking hen (i.e. an animation in which exactly the same set of dots in motion as that employed for the walking hen was presented, but with spatially randomized starting positions). Chicks tested on day 1 (experiment 2) or on day 2 (i.e. after a period in the dark following exposure on day 1 (experiment 3)) proved able to discriminate the two animation sequences: males preferentially approached the novel stimulus, females the familiar one. These results indicate that discrimination was not based on local motion vectors, but rather on the temporally integrated motion sequence.

Abstract: Recent water maze experiments suggest that rats performing place navigation primarily use the geometric information provided by a set of landmarks, and neglect the featural information provided by the identities of the landmarks. Here, I develop a model that explains how an animal may perform place navigation by relying only on geometric information. The core of the model is the representation of places as panoramas defined by circular bar-codes embodying the relative bearings and apparent sizes of the landmarks, irrespective of their identities. There are two stages in the model. During the first stage, the animal freely explores its environment in order to acquire spatial information at the local level. During the second stage, the animal uses the information previously memorized to perform place navigation towards the goal it intends to reach. The possible role of two brain areas in place navigation is discussed within this framework. Beyond their primary role in landmark-based representations of places, hippocampal place cells may be involved in computing the current distances to the landmarks. Beyond their primary role in landmark-based representations of headings, post-subicular head-direction cells may be involved in computing the “compass bearings” of the landmarks.