Unsupervised learning of perceptual feature combinations

In many situations it is behaviorally relevant for an animal to respond to co-occurrences of perceptual, possibly polymodal features, while these features alone may have no importance. Thus, it is crucial for animals to learn such feature combinations in spite of the fact that they may occur with variable intensity and occurrence frequency. Here, we present a novel unsupervised learning mechanism that is largely independent of these contingencies and allows neurons in a network to achieve specificity for different feature combinations. This is achieved by a novel correlation-based (Hebbian) learning rule, which allows for linear weight growth and which is combined with a mechanism for gradually reducing the learning rate as soon as the neuron's response becomes feature combination specific. In a set of control experiments, we show that other existing advanced learning rules cannot satisfactorily form ordered multi-feature representations. In addition, we show that networks, which use this type of learning always stabilize and converge to subsets of neurons with different feature-combination specificity. Neurons with this property may, thus, serve as an initial stage for the processing of ecologically relevant real world situations for an animal. During foraging and exploration, the neural system of animals is flooded with numerous sensory features. From this confusing signal repertoire, it needs to learn extracting relevant events often encoded by specific perceptual feature combinations. For example, a specific smell and some distinct visual attribute may be meaningful when occurring together, while by themselves these features are irrelevant. Learning this is complicated by the fact sensory signals occur with different intensity and occurrence frequency beyond the control by the animal. Here we show that it is possible to train neurons with external signals in an unsupervised way to learn responding specifically to different feature combinations largely unaffected by such presentation contingencies. This is achieved by a novel learning rule which achieves stable neuronal responses in a simple way by gradually reducing the learning rate at its synapses as soon as the neuron's response to the feature combination exceeds a certain level. This allows neurons in a network to code for different feature combinations and may facilitate ecologically meaningful evaluation of perceived situations by the animal.