Abstract
Research in decision-making has traditionally focused on understanding perceptual tasks, in which competing sets of inputs are presented simultaneously. But few studies have explored more naturalistic decisions, such as those in taste preference, where an animal samples one stimulus and has the choice to either stay or switch to another based on its intrinsic value. Given that this choice occurs across time, we ask: how do previous experiences with stimuli affect future experiences with same or different stimuli? In order to test memory’s influences on decision-making, we used Matlab to incorporate synaptic depression into an Izhikevich/LIF hybrid network modeling taste preferences in rats in two different ways: input depression and post-input depression. We compared each memory model to our control non-memory model and found that adding depression post-input created larger interactions between stimuli (post-input: n = 3453, r = -.11, p <.0001 vs. no memory: n= 4264, r = .04, p <.01), while the input depression model did not (n = 415, r = .05, p > .28). All models manipulated the distribution of time spent at stimuli. These results indicate that taste preference models, and likely other preference-related tasks, are highly susceptible to previous influences. In the future, we will incorporate other types of short and long-term memory and compare these results with experimental data to determine which model is the best biological fit.