Fig. 2: High aloof individuals exhibit reduced exploration, increased choice repetition, and altered learning dynamics.
From: Social aloofness is associated with non-social explore-exploit decisions
A Schematic of the hidden Markov model (HMM) used to label latent goal states underlying choice behaviors. The model incorporates three exploit labels for repetitions of each of the three choices, as well as an explore state in which choices were distributed across multiple options. Gray bars on observed choices represent choices labeled as exploratory. B Percentage of trials labeled as exploratory relative to aloof score. C Probability of exploring following an explore trial relative to aloof score. D Probability of exploiting following an exploit trial relative to aloof score. E Schematic of reinforcement learning model (RL) and RL choice kernel (RLCK) models tested. AIC is included to show relative model performance when including the choice history kernel. F Learning rate as measured by the RLCK model relative to aloof score. G Inverse temperature as measured by the RLCK model relative to aloof score. Higher inverse temperature indicates lower decision noise. H Choice stickiness as measured by the RLCK model relative to aloof score. Higher values indicate a stronger tendency to repeat previous choices regardless of reward. Error bars represent SEM for each bin, shaded areas represent 95% confidence interval for the regression estimate, n = 1001 for all analyses.
