The authors analyze prevalent theoretical and empirical quantitative models of choice using data from European starlings (Sturnus vulgaris). The different models originate from diverse lines of thinking, including optimal foraging theory, descriptive behavioral analysis, and temporal discounting. The authors also investigate the Sequential Choice Model (SCM) that predicts choice between simultaneous alternatives as a function of behavior in sequential encounters. SCM assumes that simultaneous encounters are rare in nature, where animals often exploit or dismiss single opportunities. Mechanisms of choice adapted for sequential encounters may be evolutionarily stable even if they impose a cost in the rarer simultaneous cases. The best predictive power is achieved by the SCM and by hyperbolic discounting with value (or attractiveness) of each option proportional to reward amount and inversely proportional to the sum of delay to reward plus delay to respond. Choice probability matched the ratio of each option's value to the sum of values of all available options. The good fit of SCM supports the view that choice is driven by mechanisms adapted to sequential, rather than simultaneous encounters with feeding opportunities.

a b s t r a c tDifferences between training and working contexts have the potential to be a major cause of deficits in performance of searching animals. Detection responses of individuals trained with high rates of target stimulus presentation tend to extinguish when moved to a new context where their rate of target encountering is low. This problem is acute with some contraband and people detection dogs where the rate of target encountering in the work context is significantly lower than during training. While the rate of extinction can be mitigated by planting known targets in the working contexts, this is often logistically difficult, dangerous, or impractical; an alternative solution would therefore be beneficial. Here, we explore the novel approach of adding non-contraband target stimuli to the training set and then presenting these innocuous targets periodically in the work context, thereby avoiding the logistic difficulties attached to the use of real contraband targets. Our rationale is that the search persistence caused by the innocuous targets could generalise to the real targets, thus increasing resistance to extinction in the latter. The potential problem with this approach is that dogs may learn to focus on the innocuous targets in the work context to the detriment of the real targets. In our experiments, 21 dogs were trained with three contraband (explosive) and one innocuous (non-explosive) odours. When they were transferred to a "work" context, they were separated into three groups, as follows: Group "0T" (zero target) were not exposed to any targets in the work environment; Group 1T (one target) were exposed to and rewarded on one innocuous target in the work environment; and Group 3T (three target) were exposed to and rewarded on three contraband targets in the work location. These regimens continued for six weeks during which time all dogs received two refresher training days away from their work location, where they were rewarded on all four target odours. Following this work phase, search and detection performance was tested in the work location for all stimuli. In the work phase, search vigilance in the 0T group dropped considerably compared with the 1T and 3T groups. Critically, when dogs were reexposed to all four targets in the work location at the end of the work phase, detection rates were significantly reduced for the 0T group, but were maintained on all targets for the 1T Abbreviations: SDD, scent detecting dogs; NE, non-explosive; DR, detection rate. 113 and 3T group. Our results show that rewarding search persistence with innocuous stimuli is potentially a successful strategy to maintain detection-dog performance across a range of trained contraband odours.Crown

Free-flying honeybees (Apis mellifera) were trained in a series of experiments designed to look for evidence of risk sensitivity in foraging for sucrose solution. The suitability of the choice method used was established in 3 preliminary experiments with differences in concentration, amount, and probability of reward. Of 5 subsequent experiments in which 2 alternatives provided the same mean concentration of sucrose solution with different variance, 3 showed risk indifference, and 2 showed risk aversion (preference for consistent reward). Of 2 final experiments in which the alternatives provided the same mean amount of sucrose solution with different variance, both showed risk aversion. Performance could be simulated quantitatively with a simple choice model developed by P. A. Couvillon and M. E. Bitterman (1991) to account for the results of a wide range of previous experiments on discriminative learning in honeybees.

SUMMARY With the exception of honeybees, there have been few good invertebrate models for associative learning. Grasshoppers and locusts (Orthoptera:Acrididae) possess a number of characteristics that make them excellent candidates for such studies, and in this paper we present a novel protocol,based on a Y-maze, that is specifically designed for studying their learning and choice behaviour. Three separate experiments were conducted using individual gregarious forms of the desert locust, Schistocerca gregaria. In our first experiment, coloured arms of a two-sided Y-maze provided a large or small amount of wheat for nine choice-trials. In the second experiment, locusts discriminated odours with wheat rewards for nine choice-trials. The odour-wheat reward combinations were then reversed for an additional nine choice-trials. For the third experiment, the locusts again discriminated odours, but here we used artificial foods and the rewards differed in their concentration of protein and digestible carbohydrate. The results indicate that, in addition to showing good acquisition of choice performance, the locusts also took less time to reach the larger-rewarded option. The data indicate that our protocol is highly sensitive for recording choice behaviour in acridids and reveals the potential they have for advancing our current understanding of associative learning and the field of learning in general.

Several factors, such as cold exposure, aging, the number of experiences and viral infection, have been shown to affect learning ability in different organisms. Wolbachia has been found worldwide as an arthropod parasite/mutualist symbiont in a wide range of species, including insects. Differing effects have been identified on physiology and behavior by Wolbachia. However, the effect of Wolbachia infection on the learning ability of their host had never previously been studied. The current study carried out to compare learning ability and memory duration in 2 strains of the parasitoid Trichogramma brassicae: 1 uninfected and 1 infected by Wolbachia. Both strains were able to associate the novel odors with the reward of an oviposition into a host egg. However, the percentage of females that responded to the experimental design and displayed an ability to learn in these conditions was higher in the uninfected strain. Memory duration was longer in uninfected wasps (23.8 and 21.4 h after conditioning with peppermint and lemon, respectively) than in infected wasps (18.9 and 16.2 h after conditioning with peppermint and lemon, respectively). Memory retention increased in response to the number of conditioning sessions in both strains, but memory retention was always shorter in the infected wasps than in the uninfected ones. Wolbachia infection may select for reduced memory retention because shorter memory induces infected wasps to disperse in new environments and avoid competition with uninfected wasps by forgetting cues related to previously visited environments, thus increasing transmission of Wolbachia in new environments.