Sensory Horizons and the Functions of Conscious Vision

"We have reviewed long-lasting postdiction effects in the 350-400ms range. These effects suggest that conscious perception does not generally occur before 350ms after stimulus onset. If conscious perception occurred before that, manipulating post-stimulus processing would be too late to change it. The fact that the fate of a percept can still be changed in that time frame indicates that it was not already conscious. This means that consciousness is slow. It is preceded by a window of unconscious sensory integration that can last up to 450ms in the most extreme cases (Herzog et al. 2020)."

What long-lasting postdiction shows, p. 11

This passage argues that conscious perception is preceded by a temporally extended window of unconscious integration, supporting the view that temporal coordination mechanisms gate entry to awareness on ~350–450 ms scales (relevant to timing and segmentation in both brains and AI systems) .

"One solution to perceptual reality monitoring (PRM) is to develop a system of higher-order representations indicating which first-order representations are reliable reflections of the world as it is now ... Under the PRM theory of consciousness, conscious representations are the result of the process of selecting reliable perceptual signals for use in the rational (model-based) control of action."

Reality monitoring, p. 25

By positing a higher-order tagging system that selects percepts for flexible control, this account links conscious access to metacognitive reportability—a key bridge between self-modeling in humans and introspection modules in AI .

Figures

Figure 4 (p. 29) : Depicts how a higher-order reliability signal can close the integration window to fix a conscious model for subsequent planning—tying metacognitive tagging to reportable conscious content and control .

"It is intriguing to speculate that the timescale of this alternation between perceptual and active inference is similar to that identified for the window of conscious perception, around 300-500ms. Recent discoveries in animals also indicate that the neural correlates of current versus hypothetical experience can alternate within a theta cycle (Kay et al., 2020)."

Computational perspectives on the timescale for consciousness and control, p. 22

Alternations between present-focused and hypothetical computation point to metastable switching—akin to state transitions in conscious access and potentially analogous to mode switches in AI models .

"The entire replayed sequence is compressed into a time window of <200ms ... Kay et al. (2020) found neural activity in the rodent hippocampus encoding two possible future scenarios ... in constant alternation at 8 Hz: one scenario per 125ms cycle."

Computational perspectives on the timescale for consciousness and control, p. 19

Sub-200 ms replay and 8 Hz alternation further support fast, discrete regime switching that could bracket or compete with conscious perceptual consolidation .

"There is an apparent tension between long-lasting parallel encoding and the apparently serial nature of the contents of experience. A recent focus in neuroscience of identifying neural activity related to a state other than the agent’s current state is prima facie evidence that the brain contains a multitude of internal processes that do not correspond to our current experience."

Consciousness and reality monitoring, p. 33

This motivates a representational selection/structuring process for conscious contents—akin to selecting a low-dimensional, coherent subspace from richer latent activity, paralleling representational geometry concerns in AI .

"Amassian et al. (1993) suppressed the perception of three letters by presenting a mask 100ms later. They then discovered that applying transcranial magnetic stimulation to the visual cortex 100ms after the mask suppressed the mask, and thus led to a recovery of the target letters. The target must have been processed unconsciously for 200ms, since a TMS pulse 200ms after its onset could still determine whether or not the target was perceived."

Postdiction and the speed of consciousness, p. 6

This causal intervention shows that specific perturbations can toggle visibility outcomes well after stimulus onset, highlighting manipulable control points for conscious access that inspire analogous ablation/patching probes in AI systems .