Active inference models of attention and self-control typically emphasize effortful precision weighting implemented through executive control. However, empirical findings indicate that attentional stability and self-regulation can improve under minimal cognitive effort. This poses a challenge for effort-centric interpretations of control cost in predictive systems [1]. We address this gap by proposing a regulatory framework in which attention optimization emerges through reductions in expected free energy mediated by brain–body coupling, rather than increased top-down control. The model reframes low-effort regulation as efficient precision allocation across interoceptive and exteroceptive hierarchies.
Methods
We formalize low-effort attention as an active inference process in which expected free energy is minimized via autonomic and cingulate-mediated precision control. The model centers on an anterior cingulate–posterior cingulate–striatal (APS) circuit interacting with parasympathetic regulation to modulate policy selection and control cost. Empirical constraints include observed shifts in midline theta–alpha dynamics, heart-rate variability, and network reconfiguration following brief low-effort training paradigms [2]. Control is modeled as adaptive precision modulation rather than sustained executive signaling.
Results
The framework predicts that reducing expected free energy can be achieved by stabilizing interoceptive predictions and lowering control-related metabolic demand. Simulated dynamics reproduce empirical signatures of low-effort regulation, including increased midline coherence, parasympathetic dominance, and transitions from frontoparietal engagement to cingulo-striatal coordination. The model accounts for rapid plasticity in cingulate pathways and the emergence of automaticity without increased policy complexity. These results suggest that attentional efficiency arises from optimized precision weighting rather than enhanced effort.
Discussion
This work situates low-effort attention and self-control within an active inference framework, offering a computational account of how regulation can improve while control cost decreases. By treating attention as precision optimization coupled to autonomic regulation, the model reconciles empirical findings with free-energy principles and generates testable predictions for electrophysiological and interoceptive markers [1, 2]. The framework generalizes across training paradigms and informs theories of embodied cognition, adaptive control, and energy-efficient artificial agents.
References
1. Tang, Y.Y., Tang, R, Posner, M. I., & Gross, J. J. (2022). Effortless training of attention and self-control: mechanisms and applications. Trends Cogn Sci. 26(7), 567-577. https://doi.org/10.1016/j.tics.2022.04.006 2. Friston, K. (2010). The free-energy principle: a unified brain theory?. Nat Rev Neurosci 11, 127–138. https://doi.org/10.1038/nrn2787
Acknowledgement
This work is supported by the ONR N000142412270 and NIH R33 AT010138.
