Learning to Synchronize: Midfrontal Theta Dynamics during Rule Switching

Abstract

In recent years, several hierarchical extensions of well-known learning algorithms have been proposed. For example, when stimulus-action mappings vary across time or context, the brain may learn two or more stimulus-action mappings in separate modules, and additionally (at a hierarchically higher level) learn to appropriately switch between those modules. However, how the brain mechanistically coordinates neural communication to implement such hierarchical learning remains unknown. Therefore, the current study tests a recent computational model that proposed how midfrontal theta oscillations implement such hierarchical learning via the principle of binding by synchrony (Sync model). More specifically, the Sync model uses bursts at theta frequency to flexibly bind appropriate task modules by synchrony. The 64-channel EEG signal was recorded while 27 human subjects (female: 21, male: 6) performed a probabilistic reversal learning task. In line with the Sync model, postfeedback theta power showed a linear relationship with negative prediction errors, but not with positive prediction errors. This relationship was especially pronounced for subjects with better behavioral fit (measured via Akaike information criterion) of the Sync model. Also consistent with Sync model simulations, theta phase-coupling between midfrontal electrodes and temporoparietal electrodes was stronger after negative feedback. Our data suggest that the brain uses theta power and synchronization for flexibly switching between task rule modules, as is useful, for example, when multiple stimulusaction mappings must be retained and used.

Theta-phase connectivity between medial prefrontal and posterior areas underlies novel instructions implementation

“…First, it might suggest that top-down synchronization is exerted similarly towards both hemispheres independently of lateralization of stimuli and responses. This interpretation is consistent with a previous study testing the model prediction in a reversal rule learning task involving lateralized responses and reporting bilateral clusters of connectivity between FCz and posterior electrodes (Verbeke et al, 2020). Information on which networks are task-relevant and which should be inhibited could be coded in other aspects of the interactive dynamics between mPFC and posterior areas, such as phase coding or cross-frequency coupling (Helfrich and Knight, 2016), or by means of activity-silent and less resource consuming neurophysiological mechanisms (Stokes, 2015; Masse et al, 2019).…”

“…According to the computational model proposed by Verguts (2017) and updated by his colleagues (Verbeke and Verguts, 2019; Senoussi et al, 2020b; Verbeke et al, 2020), theta oscillations from the mPFC control unit serve the purpose of synchronizing posterior task-relevant processing units. Therefore, we were primarily interested in investigating whether task-demands (i.e., preparing for implementation vs maintenance) affect connectivity between frontal and posterior areas.…”

Long-range cortical synchronization supports abrupt visual learning

“…Hebb’s favored mechanism for learning took the form of cell assemblies that become connected after being repeatedly activated at the same time. Modern investigations of this hypothesis in humans have confirmed that long-range synchronization between different brain regions is a reliable correlate of certain kinds of learning (Miltner, et al, 1999; Verbeke, et al, 2021). This coherence has often been found in the gamma band (Miltner, et al, 1999; Jutras, et al, 2009; Popescu, et al, 2009; Igarashi, et al, 2014), which is also associated with learning in our task (Figures 3, 6).…”

A causal role for the parietal cortex in ambiguity computations in humans

“…Also, the model included dummy regressors for the IPS-TMS stimulation and the PPC-TMS stimulation conditions and the interaction between each TMS stimulation condition and the task-related regressors. We explored frontal electrodes where theta activity related to prediction error has been described in prior work [20][21][22]30 . For both uPE regressors, we found a significant modulation (Figure 5).…”