Steve Langton

An information primacy model of exploratory and foraging behaviour

We describe a stochastic model of an animal exploring and foraging within an uncertain environment. Behaviour is determined not by an optimizing algorithm but by fuzzy systems using linguistic rules derived from the information primacy hypothesis which stresses the importance of continual information gathering under conditions of uncertainty. In the model, the animals hunger increases steadily over time and is reduced by visiting locations that may contain varying amounts of food. Uncertainty arises from three sources: (1) location novelty or ambiguity, that is, the animal is uncertain whether it has visited the same location before; (2) variation in the amounts of food in a given location; and (3) the recency of information concerning these two aspects of a given location. In complex and changing environments fresh information is likely to be more accurate than old information and consequently our model gives most weight to recently gathered information. All sources of uncertainty are reduced by visiting locations and gathering fresh information. The model is successful in simulating results from experiments investigating such phenomena as: spontaneous alternation; patrolling; the effects of hunger on the variability of learnt responses; latent learning; contrafreeloading; and behaviour following changes in food availability.

Splitting Behaviour Into Bouts; a Maximum Likelihood Approach

One method of splitting behaviour into bouts is to model the data as a mixture of two (or more) exponential distributions and to calculate a bout criterion from the resulting parameter estimates. The parameter estimates under a mixture model can be obtained using a maximum likelihood approach. The sample size required to obtain reasonable estimates of the parameters using this approach is investigated using simulated data, and found to depend on the ratio between the two densities of the two exponential processes and the proportion in which they are mixed. The use of likelihood ratio tests in helping to determine whether the data occur in bouts is also described and illustrated.

How an animal's behavioural repertoire changes in response to a changing environment: a stochastic model

We describe a stochastic model that simulates how the behavioural repertoires of animals living in complex and stochastic environments change in response to the changing environment. The model incorporates concepts derived from research in three main areas: (a) cognitive models of the environment, (b) the information-primacy hypothesis, and (c) approach/avoidance behaviours to novel and unfamiliar stimuli. The model animal is in an environment that stochastically changes, and in order to reduce primary needs it must perform behaviours that are appropriate to the changing environmental conditions. Different behaviours from its behavioural repertoire are performed and if the chosen behaviour matches the current environmental conditions the animal is able to reduce its primary needs with a probability that increases with the degree of match. A large mismatch between the performed behaviour and the environment simulates the situation where the behaviour does not meet the current environmental conditions, and hence there is a high degree of uncertainty and little chance of reducing primary needs. In such cases a plausible new behaviour may be created and added to the repertoire, and the unsuccessful behaviour can be deleted. Thus a history of small mismatch values indicates that performance of the behaviour has reliably met the environmental requirements; whilst a history of large mismatch values reflects far greater uncertainty associated with the performance of the behaviour. In the model a high priority to satisfy primary needs favours the selection of behaviours with histories of small mismatch values, and a high priority to reduce uncertainty favours behaviours with histories of high mismatch values. The degree of mismatch interacts with the current cognitive ability of the animal to process the information to create one of two types of negative affect; i.e. either boredom or anxiety. Boredom and anxiety interact with the need state to determine whether reliable or uncertain behaviours are performed. Simulations of low-variability environments result in a small behavioural repertoire, the repetitive performance of only a few behaviours, and an affective state that is neither markedly anxious nor bored. Whilst simulations of high-variability environments result in a large behavioural repertoire, the performance of many behaviours in different sequences, and an affective state that frequently shifts between high levels of boredom and anxiety. The time it takes after a switch to a novel environment for the behavioural repertoire to include behaviours successful in reducing primary needs in that novel environment is less when coming from a high-variability environment than when coming from a low-variability environment. The outputs of the model are compared to the behaviour of real animals living in environments with differing degrees of change, and the crucial role of uncertainty reduction in the creation of behavioural competence is discussed.

Cinnamamide modifies foraging behaviors of free-living birds

Chemical repellents may provide an effective and humane method of reducing bird damage to crops via modification of the feeding behavior of the target species. We observed behavior of free-living birds, in particular greenfinches (Carduelis chloris), blue tits (Parus caeruleus) and great tits (P. major), feeding on peanuts contained in wire-mesh feeders set out in 5 rows at 5-m intervals progressing away from the edge of woodland. Two identical patches of peanuts were available and were approximately 300 m apart. Prior to treatment, all birds preferred to feed closest to the woodland. We applied cinnamamide (0.6% mass/mass), an avian repellent, to peanuts in the preferred rows of 1 patch (row 1 next to the wood in the first year of the experiment, and rows 1-3 in the second year). All birds avoided treated peanuts. When row 1 was treated, the number of tits feeding on rows 2-3 increased, and many of the greenfinches moved away from the treated patch to the untreated patch. When rows 1-3 were treated, a few tits moved to feed on row 4, but most birds left the treated patch and numbers increased on the untreated patch, which suggested they flew to the untreated patch. Modifications of feeding behavior brought about by the presence of cinnamamide varied among species. Such modifications may have been related to differences in social organization: tits were relatively solitary feeders and were also probably establishing territories at the time of the experiment (Feb-Mar), whereas greenfinches fed and flew in large flocks. Thus, it was likely easier for greenfinches to fly between patches than the tits. Only when eating untreated peanuts at their local patch involved feeding ≥20 m from cover did most tits leave the patch, probably for the untreated patch. Bird pests tend to be flock feeders, thus an effective and appropriately formulated chemical repellent may be an effective tool for modifying the behavior of bird pests in order to reduce damage.