The target article by Fleming and Michel offers a thought-provoking reinterpretation of conscious vision, shifting its primary function from immediate action guidance to a more reflective role that facilitates offline cognitive processes. Their notion of “sensory horizons,” which delineates the temporal and spatial scope within which vision operates, is used to argue that the expansion of these horizons coincided with the emergence of conscious perception. According to their view, conscious vision operates slowly and deliberately, enabling integration across time and supporting planning, simulation, and complex decision-making. This framing challenges traditional action-oriented theories by positioning conscious vision as a scaffold for higher-order cognition, a view that is at once ambitious and integrative.

In this respect, the framework is perhaps conceptually too rich and leaves further important considerations unresolved. Conscious visual stimulus is often fragile and distorted, illusory in nature, subject to attentional failures and biases, all of which might at various times be considered to erode its reliability. For example, the experimental results concerning change blindness and attentional blink suggest that consciousness does not always provide a stable or accurate representation of the environment (Simons & Rensink, Reference Simons and Rensink2005). Moreover, in many cases, conscious perception is slower in temporal precision compared with its unconscious counterparts (Mudrik, Faivre, & Koch, Reference Mudrik, Faivre and Koch2014). If conscious vision evolved to support planning, how does it become so prone to errors in perception and distortions arising from within? The basis of these observations challenges the very robustness of the marvelous claim that conscious vision is primarily an adaptation for deliberation rather than an instability in representation.

The authors also link the evolution of consciousness to improvements in reality monitoring, the ability to discriminate internally generated representations from external inputs. Yet reality monitoring is a fragile and fallible capacity, easily disrupted in both everyday contexts and psychopathology. Disorders such as schizophrenia, PTSD, and dissociative states provide clear evidence of its breakdown, often in individuals who retain intact sensory horizons (Stephan-Otto et al., Reference Stephan-Otto, Núñez, Lombardini, Cambra-Martí, Ochoa, Senior and Brébion2023). This disconnect challenges the causal chain proposed by Fleming and Michel: expanding sensory horizons does not guarantee reliable reality monitoring. Contemporary neuroscientific frameworks, including predictive coding and active inference, highlight that reality monitoring depends less on the scope of sensory input and more on the precision weighting of hierarchical predictions, which is modulated by attention, prior beliefs, and context (Laukkonen et al., Reference Laukkonen, Friston and Chandaria2025). From this perspective, failures in reality monitoring stem from dysregulated priors or metacognitive misalignment, not simply from limited sensory horizons.

One should note that this limitation becomes more relevant when one tries to apply it to clinical psychology. Numerous psychiatric conditions can be thought of as disruptions of affordance perception and reality monitoring. Individuals with major depressive disorder fail to perceive positive affordances and view neutral or pleasant stimuli as irrelevant or threatening (Disner et al., Reference Disner, Beevers, Haigh and Beck2017). Those with social anxiety disorder perceive threats in benign social situations, and PTSD patients are hypervigilant to even purportedly safe cues of danger (Vanhollebeke et al., Reference Vanhollebeke, De Smet, De Raedt, Baeken, van Mierlo and Vanderhasselt2022). In schizophrenia, breaches of source-monitoring induce hallucinations and delusions, where internally generated content is experienced as externally real (Mondino et al., Reference Mondino, Dondé, Lavallé, Haesebaert and Brunelin2019). So, these disorders portray reality monitoring as fragile and adversely question the ubiquitous assumption that consciousness is equated with the reliable discrimination of inner and outer signals.

Neuroscience presents convergent evidence for the presence of neural correlates for these distortions. Hyperamygdala activity and inferior vmPFC functioning define an anxious disorder, contributing to the indefinite perception of a threatening event (Shin & Liberzon, Reference Shin and Liberzon2022). Depression may bring on attenuation of dorsolateral prefrontal cortex (DLPFC) function and faulty default mode network (DMN) connectivity that diminishes the vicissitude of attentional shifting toward positive cues (Drysdale et al., Reference Drysdale, Grosenick, Downar, Dunlop, Mansouri, Meng and Liston2023). The hippocampal dysfunction in PTSD impairs contextual memory of safety and differentiation of safe context, wherein a person who is supposed to differentiate a safe environment feels threatened (Bisby et al., Reference Bisby, Horner, Bush and Burgess2020). Such findings support the view wherein the reality-monitoring failures described clinically are rooted in neurobiological dysregulation, giving any all-encompassing theory on conscious vision the task of working integratively with such evidence.

So, the essence of these points is to highlight what I see as a gap in Fleming and Michel’s model: While it integrates evolutionary and philosophical formations pretty well, it runs short regarding the variability and fragility of conscious vision in clinical and neurobiological arenas. Consciousness is not merely a stable window for planning; it is rather an evolving interface, shaped by predictive processes, affective states, and neural constraints. To further enrich their framework, incorporating clinical evidence and neuroscientific data about the breakdown of conscious vision and reality monitoring in pathological conditions would prove fruitful. This would bring the model into the realm of abnormal consciousness and test it against assessments of ordinary consciousness in all its facets.

This criticism also offers several avenues for empirical exploration. If conscious vision and reality monitoring do presuppose specific neural circuits, then neuromodulation techniques can be harnessed to investigate such circuits and confirm their causal role. For example, one could intervene by applying rTMS to the DLPFC or vmPFC to find out if modulating these regions in subjects from clinical groups changes affordance perception and reality monitoring. An initial wave of clinical research is showing that rTMS of the DLPFC somehow improves mood and attentional flexibility in depression (O’Reardon et al., Reference O’Reardon, Solvason, Janicak, Sampson, Isenberg, Nahas and Sackeim2022), raising the possibility that such interventions might also force affordance perception to the straight and narrow. Similarly, silencing hyperactive threat circuits in anxiety could possibly dampen the exaggerated detection of danger. Such translational actigraphy, on the one hand, would test the validity of Fleming and Michel’s framework, and on the other hand, lay down the first bricks for clinical applicability.

In conclusion, Fleming and Michel’s article provides a bold and integrative proposal for understanding conscious vision as a scaffold for offline cognition, anchored in the notion of sensory horizons. Yet by underestimating the fragility of conscious perception and frequent failures in reality monitoring, the framework risks overstating evolutionary reliability. Integrating evidence from clinical psychology and neuroscience would show that conscious vision is dynamic, predictive, and often unreliable. A future model incorporating these complications would align more closely with empirical data and have stronger translational relevance for understanding and addressing perceptual and consciousness disorders.