Best model: hyperbolic model $SV_t = reward_t \times \frac{1}{1 + k * effort_t}$ ```r > bic_summed model aic bic n 1: hyperbolic 19017.25 19587.43 77 # lowest summed BIC across subjects 2: linear 19286.52 19856.69 77 3: sigmoid 19089.12 19944.38 77 4: parabolic 19416.46 19986.63 77 > dt1[, .(aic = mean(aic), bic = mean(bic)), by = .(model)] %>% + arrange(bic, aic) %>% data.table() # %>% fwrite("bic_mean.csv") model aic bic 1: hyperbolic 246.9773 254.3822 # lowest mean BIC across subjects 2: linear 250.4742 257.8791 3: sigmoid 247.9107 259.0179 4: parabolic 252.1618 259.5666 ``` ![[summed_bic 1.jpg]] ```r > dt2_avg model N N_prop 1: linear 7 0.09090909 # best model for 7 subjects 2: hyperbolic 47 0.61038961 # best model for 47 subjects 3: sigmoid 10 0.12987013 4: parabolic 13 0.16883117 ``` ![[model_count 5.jpg]]