Seth (this issue) does an admirable job of integrating recent theories in cognitive science. His proposal resonates with enactive theory, which also builds on counterfactual interactions. Beaton (2013) uses this concept to accommodate cases of non-veridical experiences and locked-in syndrome; Kyselo and Di Paolo (in press) suggest that any world-directed activity “also implies some sort of virtual bodily activity.” What is perceived is shaped by its meaning for the agent and her potential actions. Predictive processing arrives at a similar insight, since it suggests that “perception is inextricably tied up not just with action but also with something functionally akin to imagination” (Clark, 2012, p. 761). Enactive theory is therefore able to resolve the challenge posed by synesthesia in a way reminiscent of Seth’s proposal. But there are also important differences.

For instance, while predictive processing elevates the reduction of uncertainty to a unifying principle of cognitive science, this is unacceptable to enactive theory (Froese & Ikegami, 2013). Living is theorized to be a process of sense-making, and meaning is embodied in the organism’s precarious existence. On this view, uncertainty is a basic property of mind, while Seth has to make an exception for synesthetic concurrents. Their constancy is in tension with a theory that argues for the elimination of counterfactuals leading to sensory prediction errors.

Enactive theory also rejects the internalism and the representationalism of predictive processing (Froese & Ikegami, 2013). Phenomena such as synesthesia make it evident that the former has to pay more attention to neural dynamics, but this concession does not equate to an acceptance of internalism. There is still the disjunctivist alternative, which holds that “in those cases in which are not misled, the layout of reality shapes the contours of our conscious experiences” (Fish, 2009, p. 181). The brain is certainly an essential part of human living, but it is a mediating organ and not an isolated “black box” (Clark, 2012, p. 759). This alternative stance is not just cosmetic, as I will now try to indicate.

Seth argues that synesthetic concurrents lack perceptual presence because they are associated with a comparatively smaller space of counterfactual interactions. This may be true, but not in the way Seth intends. He conflates two independent phenomenological properties of our experience: He equates perceptual presence (the appearance of objecthood) with subjective veridicality (the appearance of realness). These properties can coincide, as in Macbeth’s soliloquy about seeing a dagger, whose unreality was only revealed by its unexpected intangibility. But many hallucinated objects do not appear as being real, thereby requiring us to discriminate objecthood and realness. Seth comes close to this distinction when he notes that concurrents are “perceived as unreal and as lacking in external object-hood” (emphasis added).

The problem is that Seth’s counterfactual predictive processing only addresses the appearance of objecthood, not of realness. But it is the absence of the latter, not of the former, that is an essential property of synesthetic experience. While many forms of synesthesia only evoke sensations that lack objecthood, such as seeing extra colors, this is not always the case. Consider this verbal report of taste-object synesthesia: “With an intense flavor […] the feeling sweeps down my arm into my hand and I feel shape, weight, texture, and temperature as if I’m actually grasping something” (Cytowic & Eagleman, 2009, p. 3). Similarly, the phenomenology of spatial sequence synesthesia includes objects such as tunnels, yet this contextual space appears as distinct from the real world (Gould, Froese, Barrett, Ward, & Seth, in preparation). A relative poverty of counterfactuals helps to explain the absence of objecthood in some forms of synesthesia, but by itself it cannot explain why concurrents are always experienced as non-veridical.

It is in this regard that an enactive account of perception offers advantages. It does not need to explain why veridical experiences appear as real on the basis of their internal representations. Instead, an object is experienced as real during veridical perception precisely because that experience itself is shaped directly by the real world (Beaton, 2013; Fish, 2009). This constitutes the phenomenology of its observer-independence, its overflowing phenomenal richness, and hence its open-ended capacity to surprise us. Conversely, when experiencing an object with which we are not directly in contact, i.e., which is non-veridical, the phenomenology of its reality is absent by default. An imaginary object’s lack of subjective veridicality does not require an additional explanation. To be sure, delusions of doxastic veridicality regarding non-veridical experiences demand a more complicated story, but, as Seth acknowledges, the transition to the latter also presents an interesting challenge to predictive processing.