| Foraging Through Prediction (2007) | |||||||||||||||
Abstract | |||||||||||||||
| To survive, an animal must use sensory events to predict the presence of mates, food, danger, and various other stimuli that are important for its survival and procreation. Although reliable prediction is critical, it is not understood how such prediction is carried out by nervous systems. We present a model which utilizes diffuse neuromodulatory systems to implement a predictive version of a Hebbian rule, and embed this rule in a feasible neural architecture. The predictive model suggests a unified way in which neuromodulatory influences are used to bias actions and control learning. When required to forage in a stochastic environment, the model captures the strategies seen in the behavior of bees and a number of other animals. It further suggests that predictive rules for synaptic plasticity offer a simple framework which is nevertheless more powerful than correlational accounts. Introduction Any animal presented with a real environment must have a means to react adaptively to that ... | |||||||||||||||
Publication details | |||||||||||||||
| |||||||||||||||