Table 2. Model parameters, priors, hyperparameters and hyperpriors.

Parameter	Prior	Hyperparameters	Hyperpriors
prior mean, $R_0^{\tau s}$	$R_0^{\tau s}\sim$ Gaussian(${\mu }_{R_0}^{\tau }$, ${\sigma }_{R_0}^{\tau }$)	${\theta }_{R_0}^{\tau }=({\mu }_{R_0}^{\tau },{\sigma }_{R_0}^{\tau })$	${\mu }_{R_0}^{\tau }\sim$ Gaussian( 50, 14 ) ${\sigma }_{R_0}^{\tau }\sim$ Gamma( 1, 0.001 )
initial learning rate, ${\alpha }_1^{\tau s}$	${\alpha }_1^{\tau s}\sim$ Beta($a_{{\alpha }_1}^{\tau }$, $b_{{\alpha }_1}^{\tau }$)	${\theta }_{{\alpha }_1}^{\tau }=(a_{{\alpha }_1}^{\tau },b_{{\alpha }_1}^{\tau })$	$a_{{\alpha }_1}^{\tau }\sim$ Uniform( 0.1, 10 ) $b_{{\alpha }_1}^{\tau }\sim$ Uniform( 0.5, 10 )
asymptotic learning rate, ${\alpha }_{\mathrm{\infty }}^{\tau s}$	${\alpha }_{\mathrm{\infty }}^{\tau s}\sim$ Beta($a_{{\alpha }_{\mathrm{\infty }}}^{\tau }$, $b_{{\alpha }_{\mathrm{\infty }}}^{\tau }$)	${\theta }_{{\alpha }_{\mathrm{\infty }}}^{\tau }=(a_{{\alpha }_{\mathrm{\infty }}}^{\tau },b_{{\alpha }_{\mathrm{\infty }}}^{\tau })$	$a_{{\alpha }_{\mathrm{\infty }}}^{\tau }\sim$ Uniform( 0.1, 10 ) $b_{{\alpha }_{\mathrm{\infty }}}^{\tau }\sim$ Uniform( 0.1, 10 )
information bonus, $A^{\tau shu}$	$A^{\tau shu}\sim$ Gaussian(${\mu }_A^{\tau hu}$, ${\sigma }_A^{\tau hu}$)	${\theta }_A^{\tau hu}=({\mu }_A^{\tau hu},{\sigma }_A^{\tau hu})$	${\mu }_A^{\tau hu}\sim$ Gaussian( 0, 100 ) ${\sigma }_A^{\tau hu}\sim$ Gamma( 1, 0.001 )
spatial bias, $B^{\tau shu}$	$B^{\tau shu}\sim$ Gaussian(${\mu }_B^{\tau hu}$, ${\sigma }_B^{\tau hu}$)	${\theta }_B^{\tau hu}=({\mu }_B^{\tau hu},{\sigma }_B^{\tau hu})$	${\mu }_B^{\tau hu}\sim$ Gaussian( 0, 100 ) ${\sigma }_B^{\tau hu}\sim$ Gamma( 1, 0.001 )
decision noise, ${\sigma }^{\tau shu}$	${\sigma }^{\tau shu}\sim$ Gamma($k_{\sigma }^{\tau hu}$, ${\lambda }_{\sigma }^{\tau hu}$)	${\theta }_{\sigma }^{\tau hu}=(k_{\sigma }^{\tau hu},{\lambda }_{\sigma }^{\tau hu})$	$k_{\sigma }^{\tau hu}\sim$ Exp( 0.1 ) ${\lambda }_{\sigma }^{\tau hu}\sim$ Exp( 10 )