The paper by Feldman and Friston explores the relationship between attention, uncertainty, and free-energy in the context of hierarchical perception. They propose that attention can be understood as inferring the level of uncertainty or precision during perception. Using neuronal simulations, they demonstrate how attention can be explained through the optimization of free-energy, which bounds surprise or the negative log-evidence for internal models of the world. The simulations assume that neuronal activity encodes a probabilistic representation of the world that optimizes free-energy in a Bayesian manner. The authors show that if the precision of sensory data depends on the states of the world, this can explain various aspects of attention, such as attentional bias, competition for attentional resources, and speed-accuracy trade-offs. They illustrate these points using the Posner paradigm, where both attended and non-attended stimuli are presented simultaneously, leading to biased competition for neuronal representation. The paper also discusses the neurobiological mechanisms underlying attention, including the role of gamma oscillations and cholinergic neurotransmission in modulating synaptic gain. Overall, the authors argue that attention is a process of optimizing synaptic gain to represent the precision of sensory information during hierarchical inference, and that state-dependent changes in precision can be modeled in the brain through activity-dependent modulation of synaptic gain.The paper by Feldman and Friston explores the relationship between attention, uncertainty, and free-energy in the context of hierarchical perception. They propose that attention can be understood as inferring the level of uncertainty or precision during perception. Using neuronal simulations, they demonstrate how attention can be explained through the optimization of free-energy, which bounds surprise or the negative log-evidence for internal models of the world. The simulations assume that neuronal activity encodes a probabilistic representation of the world that optimizes free-energy in a Bayesian manner. The authors show that if the precision of sensory data depends on the states of the world, this can explain various aspects of attention, such as attentional bias, competition for attentional resources, and speed-accuracy trade-offs. They illustrate these points using the Posner paradigm, where both attended and non-attended stimuli are presented simultaneously, leading to biased competition for neuronal representation. The paper also discusses the neurobiological mechanisms underlying attention, including the role of gamma oscillations and cholinergic neurotransmission in modulating synaptic gain. Overall, the authors argue that attention is a process of optimizing synaptic gain to represent the precision of sensory information during hierarchical inference, and that state-dependent changes in precision can be modeled in the brain through activity-dependent modulation of synaptic gain.