Back

Social Curiosity in Monkeys

Joiner, J. A.; Nair, A. R.; Fagan, N. A.; Chang, S. W. C.

2022-11-15 animal behavior and cognition
10.1101/2022.11.15.516657 bioRxiv
Show abstract

Humans and nonhuman animals derive value from many different sources. Some of these sources, notably primary reinforcers like food, are clearly rewarding and can powerfully shape our behaviors. Other sources of reward, such as information, are more intangible and abstract. Like humans, nonhuman primates also find information inherently rewarding, displaying a preference to reveal information about upcoming primary reinforcers even though the information has no bearing on the outcome itself. For animals living in social groups, the need for information extends beyond primary reinforcers like food. They need to acquire information from others or about others. This is especially true for animals, such as nonhuman primates, which live in large and often hierarchically organized societies, where processing social information for the purpose of learning and socializing can be just as critical to survival. To quantify curiosity for knowing social information in monkeys, we extend the advanced informationseeking paradigm (Bromberg-Martin and Hikosaka, 2009) into the realm of abstract, social information. We first replicated the finding that monkeys prefer advanced information about juice size (Bromberg-Martin and Hikosaka, 2009). We next trained monkeys on a social variant of this task. In the advanced social information-seeking task, monkeys had the option to choose a cue that tells them in advance the valence, or facial expression, on a monkeys face that they will be viewing before receiving the invariant amount of juice. Even though this cue did not impact which facial expression the monkeys would see, they preferred to know the valence of the facial expression in advance. Our results indicate that information-seeking behavior generalizes to seeking social information. Our findings also suggest that curiosity in nonhuman primates can be translated into increasingly abstract levels of information.

Matching journals

The top 5 journals account for 50% of the predicted probability mass.

50% of probability mass above

"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.