A Network Activity Reconfiguration Underlies the Transition from Goal to Action

Abstract

Graphical AbstractHighlights d A minority of PF neurons code the goal during both working memory and action selection d Those neurons change or maintain goal preference with equal probability d A network of heterogeneous cell assemblies can explain the observed dynamics SUMMARY Neurons in prefrontal cortex (PF) represent mnemonic information about current goals until the action can be selected and executed. However, the neuronal dynamics underlying the transition from goal into specific actions are poorly understood.Here, we show that the goal-coding PF network is dynamically reconfigured from mnemonic to action selection states and that such reconfiguration is mediated by cell assemblies with heterogeneous excitability. We recorded neuronal activity from PF while monkeys selected their actions on the basis of memorized goals. Many PF neurons encoded the goal, but only a minority of them did so across both memory retention and action selection stages. Interestingly, about half of this minority of neurons switched their goal preference across the goal-action transition. Our computational model led us to propose a PF network composed of heterogeneous cell assemblies with single-state and bistable local dynamics able to produce both dynamical stability and input susceptibility simultaneously.

Instability with a purpose: how the visual brain makes decisions in a volatile world

“…Evidence for metastable attractor dynamics in cortical activity is accumulating steadily [37,[73][74][75][76]99]. Distinct activity states with exponentially distributed durations have been reported in sensory cortex [37,75], consistent with noise-driven escape transitions [36,47].…”

Slower prefrontal metastable dynamics during deliberation predicts error trials in a distance discrimination task

“…The role of prefrontal cortex in decision making has been demonstrated using a variety of tasks (Kim and Shadlen, 1999; Hoshi et al, 2000; Freedman et al, 2001; Tanji and Hoshi, 2001; Genovesio et al, 2012; Maoz et al, 2013; Marcos and Genovesio, 2016; Rich et al, 2018), including tasks with decisions in the temporal (Genovesio et al, 2009) and spatial (Genovesio et al, 2011) domains. In particular, we have shown previously that prefrontal neurons encode the decision about which stimulus was presented farther from the center in a distance discrimination task (Genovesio et al, 2011), as well as the transformation of goals into action (Marcos et al, 2019). Prefrontal neurons also code for the duration and distance magnitude to be maintained in working memory using domain-specific representations of distance and duration in PFdl (Marcos et al, 2017).…”

Minimally dependent activity subspaces for working memory and motor preparation in the lateral prefrontal cortex

“…Complex flexible behaviors require the integration of multiple types of information, including information about sensory properties, task rules, items held in memory, items being attended, actions being planned, and rewards being expected, among others. A large proportion of neurons in the lateral prefrontal cortex (LPFC) encode a mixture of two or more of these types of information ( Rigotti et al, 2013 ; Parthasarathy et al, 2017 ; Masse et al, 2019 ; van Ede et al, 2019 ; Marcos et al, 2019 ). This mixed selectivity endows the LPFC with a high-dimensional representational space ( Rigotti et al, 2013 ), but it also presents the challenge of understanding how downstream regions that receive mixed-selective input from the LPFC can read out meaningful information.…”