Speakers
Description
When animals forage together, they should remain alert to others and environmental uncertainties. In monkeys, this commonly embodies as looking at others and around while manually foraging. We asked how do monkeys navigate the trade-off between maintaining vigilance and acquiring food? To address this question, we designed a dyadic foraging task for freely moving rhesus macaques in which they were free to search for food on a floor grid of woodchip piles. We quantified three types of actions: foraging, looking at the partner, and vigilance scan, defined as a side-to-side head movement while looking up. By computing an action transition matrix, we found that foraging and vigilance scans often follow block-like patterns as each of them occurs repeatedly before the animal performs any of the two other actions. This raises the question of whether the high probability of an action repeating is due to the cost of switching to other actions or does it reflect the influence of a sustained internal state that biases behavior? We employed a Hidden Markov Model in which the emitting actions in each hidden state are predicted using a generalized linear model (GLM-HMM). The fitted HMM-GLM revealed that states were stable, with a higher probability of staying in a state than switching. While in one state, the probability of foraging was the highest, in another, looking at the partner was most probable. These results suggest that our model was able to infer hidden states of vigilance, influencing the trade-off between maintaining vigilance and acquiring food.
