Decision making in a dynamically structured holographic memory model: Learning from delayed feedback

We present a first step towards developing a cognitive model of decision making using Dynamically Structured Holographic Memory (DSHM). DSHM is a class of memory models used to model human performance in game play, memory tasks, and language learning. The ability to detect and predict patterns is an integral part of memory models and an important part of decision making. However, decision making also requires the ability to evaluate states as more or less desirous in order to motivate the decisions. We apply a DSHM model to the decision-making task of Walsh and Anderson (2011). By initializing memory to a state of optimism and by making the model sensitive to dependencies between non-consecutive events, we find that the model is able to learn the task at a rate similar to humans.