Abstract: Under the conventional assumption that natural selection favors choice behavior that maximizes some fitness-related currency, a forager making repeated binary choices should consistently choose the more valuable option. Under the alternative assumption that natural selection favors choice behavior that minimizes costly errors, erroneous choice is not only expected but is expected to be common when the cost of errors is low. This cost depends on the potential rate of return: the higher this rate, the smaller the cost of choosing the less valuable option. When this rate is very high, a forager may err frequently and yet forego no appreciable fitness gain. Errors should thus be more common when interruptions to foraging are shorter. Our experimental results supported this prediction: gray jays chose the less valuable option more frequently when subjected to shorter interruptions (experimentally imposed delays to access to food rewards). This tendency is consistent with the idea that an adaptive decision-making process may routinely produce errors, not because errors are in some way adaptive but because their fitness cost is minimal, particularly when delays are short. From a proximate perspective, this tendency to commit errors more frequently following shorter delays may be due to constraints on the jays' information-processing capacity. In general, choice behavior should be viewed as the joint byproduct of adaptive decision making and cognitive constraints.

Abstract: This study investigated whether dolphins would show evidence of equivalence class formation between auditory stimuli. Bottlenose dolphins were trained to press one or other of two response levers depending on which one of four auditory stimuli had been previously presented. Once they had learned the initial discriminations, the stimulus-lever contingencies was repeatedly reversed. Within any given session, however, pressing of one lever always led to reward with one set of two tones and pressing the other lever led to non-reward with an alternative set of two tones. After sufficient experience with this response reversal procedure, the dolphins spontaneously chose the same levers they had first learned to be correct with one of the across-set stimulus pairs when later in the session they were presented with the other of the across-set stimulus pairs. They thus demonstrated that they had associated the tones belonging to the two sets within two separate functional classes. It is discussed why the dolphins succeeded with auditory stimuli when they had previously failed in a similar task with visual stimuli.

Abstract: Five-day-old chicks were accustomed to follow an imprinted object (a small red ball with which they had been reared) that was moving slowly in a large arena, until it disappeared behind an opaque screen. In experiments, each chick was initially confined in a transparent cage, from where it could see and track the ball while it moved towards, and then beyond, one of two screens. The screens could be either identical or differ in colour and pattern. Either immediately after the disappearance of the ball, or with a certain delay, the chick was released and allowed to search for its imprinted object behind either screen. The results showed that chicks took into account the directional cue provided by the ball movement and its concealment, up to a delay period of about 180 s, independently of the perceptual characteristics of the two screens. If an opaque partition was positioned in front of the transparent cage immediately after the ball had disappeared, so that, throughout the delay, neither the goal-object nor the two screens were visible, chicks were still capable of remembering and choosing the correct screen, though over a much shorter period of about 60 s. The results suggest that, at least in this precocial bird species, very young chicks can maintain some form of representation of the location where a social partner was last seen, and are also capable of continuously updating this representation so as to take into account successive displacements of the goal-object.

Abstract: Turtles (Pseudemys scripta) were trained in place, cue and control open-field procedures. The turtles trained in both the place and the cue procedures were able to learn their respective tasks with accuracy. Subsequent probe tests revealed that the turtles trained in the place task relied on the information provided by the extramaze cues to locate the goal. However, for these animals, no single cue was essential for performance, as accurate navigation to the goal was still possible when subsets of extramaze cues were eliminated. Furthermore, the turtles trained in the place task were able to navigate accurately to the goal place from new start locations. These results suggest that the turtles trained in the place task used map-like, relational strategies, by encoding the simultaneous spatial relationships between the goal and the extramaze cues in an allocentric frame of reference. In contrast, the turtles trained in the cue procedure used guidance strategies, i.e. approaching the individual intramaze cue associated to the goal as it were a beacon and largely ignoring the extramaze cues. Thus, the results of this experiment suggest that turtles are able to employ spatial strategies that closely parallel those described in mammals and birds.

Abstract: Two experiments investigated spontaneous class grouping behavior by human-enculturated and language-reared bonobos (Pan paniscus) and common chimpanzees (P. troglodytes). In experiment 1, three chimpanzees ranging in age from 6 to 18 years were presented with six objects. The objects embodied three conditions: additive, multiplicative and disjoint classes. All chimpanzees spontaneously produced single- and two-category classifying. In experiment 2, six chimpanzees ranging in age from 6 to 21 years were presented with 12 objects in the same class conditions. Chimpanzees mainly produced single-category classifying. Their two-category classifying was more rudimentary than that found in experiment 1. Chimpanzees did not produce any three-category classifying which would be necessary to construct the hierarchies that humans begin to construct during early childhood.