Decoding Hierarchical Control of Sequential Behavior in Oscillatory EEG Activity

Abstract

Whether we dance or compose music, write computer code, or plan a speech: A limited number of basic elements--dance moves, notes, code segments, or arguments--need to be combined in a goal-appropriate order. Following Lashley (1951), many theorists believe that such sequential skills are achieved through a set of abstract, content-independent control representations that code the position of basic elements within shorter subsequences (i.e., chunks) or the position of chunks within a hierarchically organized plan ). Empirically, it is exactly the abstract nature of serial-order control representations that has made it difficult to provide direct evidence for their existence, to distinguish them from chaining-based representations, or to characterize their functional properties.The current work tests the general hypothesis that the pattern of frequency-band specific rhythmic activity in the EEG reflects different types of serial-order control representations and allows tracking the current position within a hierarchical control structure. This hypothesis is based on two sets of observations.

Distinct Oscillatory Dynamics Underlie Different Components of Hierarchical Cognitive Control

“…Our findings suggest a relationship between theta oscillations and set-size, and this finding is consistent with previous studies that reported theta oscillations scale with working memory load (Jensen and Tesche 2002, Meltzer, Negishi et al 2007, So, Wong et al 2017, Berger, Griesmayr et al 2019. Other studies have also found that theta oscillations (presumably from frontal cortex) increase during tasks that required cognitive control (Cohen 2011, Hsieh, Ekstrom et al 2011, Kikumoto and Mayr 2018. Theta-gamma coupling has been suggested as a mechanism by which multiple representations are organized for working memory (Bahramisharif, Jensen et al 2018) and long-term memory (Heusser, Poeppel et al 2016).…”

Distinct neural representations of content and ordinal structure in auditory sequence memory

“…Here we focused on the ordinal information of a tone sequence, a typical and important relational structure to sort contents in auditory experience. Indeed, neural coding of ordinal position has been found in both animal recordings (Fortin et al, 2002;Naya & Suzuki, 2011) and neuroimaging studies (Baldassano et al, 2017;Carpenter et al, 2018;DuBrow & Davachi, 2016;Hsieh et al, 2014;Kikumoto & Mayr, 2018;Liu et al, 2020;Rajji et al, 2017;Roberts et al, 2013;Huang et al, 2018). Our results are thus consistent with previous findings and further expand to auditory sequence memory, particularly during the maintaining period.…”

Robot navigation as hierarchical active inference

“…In rodent studies, H/E couples with multimodal (hexagonally organized) value representations of the ventromedial prefrontal cortex [31,32], which may not only allow navigation to be organized according to a "common neural currency" of expected organismic value, but may also aid in updating those estimates based on histories of experience organized according to spatiotemporal trajectories. Similar bidirectional functional relationships may also be observed with the dorsomedial prefrontal cortex (and corresponding posterior structures), both allowing H/E successor representations to be informed based on activity in high level motor and gaze direction, but also potentially allowing visual and somatic spatial fields to be directed as a kind of navigation and foraging through informational and affordance spaces [33,34], with saccades/samples and discrete actions orchestrated at theta frequencies [35,36]. Somewhat less exotically, a substantial amount of predictive transitions between place fields may be explainable by relatively simple "bump attractor" models of CA3 [37,38], or in terms of the ability of recurrent networks to encode predictive informationas a kind of spontaneous meta-learning -via their evolving attractor dynamics [39].…”