Nijhawan makes a compelling case that the compensation for neural delays must occur at some level, and possibly many levels, of the human central nervous system. In the several sections of this argument, he mentions research from prism adaptation. Prism adaptation is clear evidence that the human central nervous system compensates for its own internal spatial characteristics. If the sensorimotor system can compensate for internal spatial characteristics, might it also be able to compensate for internal temporal characteristics (i.e., neural delays)? As Nijhawan points out, it must, since failure to do so would yield an organism that cannot interact well with the world.

The field of prism adaptation starts with the observation that intersensory spatial arrangements are not constant over the course an organism's life, but change both slowly (e.g., during maturation) and rapidly (e.g., when reaching through water). Any organism that cannot adapt its compensatory mechanisms to deal with changes in intersensory relationships will have nearly as much difficulty interacting with the world as an organism that has no compensatory mechanisms. Prism adaptation shows that human sensorimotor systems can adjust to both rapid and slow changes in intersensory relationships (for reviews, see Bedford Reference Bedford1993; Welch Reference Welch1978).

Neural delays also change both rapidly and slowly during each organism's lifetime (see, e.g., Ahissar & Ahissar Reference Ahissar and Ahissar1994). Thus, if humans can compensate for the temporal characteristics of the sensory systems like they do for the spatial characteristics, might not this mechanism also be able flexible enough to respond to changes in neural delay? Since the 1950s, several researchers have shown that this does not seem to be the case (see, e.g., Sheridan & Ferrel Reference Sheridan and Ferrel1963; Smith et al. Reference Smith, McCrary and Smith1962; Reference Smith, Wargo, Jones and Smith1963). It has, however, since been conclusively demonstrated that not only do humans compensate for external temporal shifts, but the mechanism for doing so is remarkably similar to that involved in prism adaptation (Cunningham et al. Reference Cunningham, Billock and Tsou2001a; Reference Cunningham, Chatziasros, von, Heyde and Bülthof2001b). In general, the introduction of an external visual delay initially impairs performance, but a small amount of practice returns behavior to nearly normal levels, and the adapted state generalizes to similar situations. Subsequent removal of the delay produces a large renewed drop in performance as the newly adopted state of temporal compensation ceases to be appropriate (this negative aftereffect is the hallmark of the semipermanent nature of sensorimotor adaptation). Subsequent work has confirmed this effect (Fajen Reference Fajen2007; Fujisaki et al. Reference Fujisaki, Shimojo, Kashino and Nishida2004; Miall & Jackson Reference Miall and Jackson2006; Navarra et al. Reference Navarra, Soto-Faraco and Spence2007; Stetson et al. Reference Stetson, Cui, Montague and Eagleman2006; Vatakis et al. 2004; Reference Vatakis, Navarra, Soto-Faraco and Spence2007). It is critical here to note that, unlike the visual prediction effect speculated to exist for moving objects, the temporal adaptation effect cannot be completely explained by spatial extrapolation for many reasons, including the fact that it occurs for discrete, stationary stimuli.

Nijhawan focuses explicitly on compensating for continuous events, with special emphasis on moving objects. There must, however, be some form of temporal compensation that also works for discrete events. The reason for this is simple: neural latencies differ drastically for different sensory modalities as well as within a modality for different stimuli (e.g., Bolz et al. Reference Bolz, Rosner and Wässle1982; Sestokas & Lehmkuhle Reference Sestokas and Lehmkuhle1986). The absence of a coordinating influence that can also compensate for brief, non-moving events would necessarily lead to a fragmentation of the phenomenally experienced world. For example, the brighter portions of a briefly seen object would seem to have occurred at different time than the darker portions. As multimodal, briefly presented, non-moving objects are not perceptually fragmented, and because many of the demonstrations of temporal adaptation have successfully used stationary, briefly presented, discrete stimuli, there must be some form of temporal compensation that does not involve spatial prediction.

Finally, in their experiments, Cunningham et al. (Reference Cunningham, Billock and Tsou2001a; Reference Cunningham, Chatziasros, von, Heyde and Bülthof2001b) noted that many participants complained about apparent perceptual changes during temporal adaptation. More specifically, as behavior in the delayed conditions approached normal levels, the participants began to indicate that the visual and proprioceptive events, which were separated by up to 400 msec, seemed to occur simultaneously. A number of studies have since empirically confirmed this perceptual shift in the point of subjective simultaneity (Fujisaki et al. Reference Fujisaki, Shimojo, Kashino and Nishida2004; Navarra et al. Reference Navarra, Soto-Faraco and Spence2007; Vatakis et al. Reference Vatakis, Navarra, Soto-Faraco and Spence2007; Vroomen et al. Reference Vroomen, Keetels, de Gelder and Bertelson2004). Several of these studies used brief, discrete, stationary stimuli. In general, these results suggest that adaptation to temporal delays is at least partially perceptual in nature. Nijhawan argues that prism adaptation, to the degree that it occurs in the visual system, is support for the theory that the visual system is engaged in prediction. Thus, the evidence that humans can both behaviorally and perceptually compensate for changes in delayed information would seem to be support for the claim of visual prediction in general, if not for spatial extrapolation in particular.

In sum, the human sensorimotor systems can compensate not only for their own internal characteristics, but also for external changes in both the spatial and temporal relationships between the senses. Moreover, at least for temporal adaptation, this compensation occurs partially in the perceptual systems, implying that they are engaging in some form of prediction. As temporal adaptation also works for discrete as well as continuous events and is not necessarily motion based, it cannot be fully explained by spatial extrapolation. Nijhawan states that, “The goal of visual prediction is to use priors contained in the unfolding visual stimulus to create a perceived state of the world that matches, as far as possible, the actual state of the world” (sect. 4, para. 2). Given what is known about how spatial and temporal compensation mechanisms adapt, this statement might be amended to make the strong involvement of previous experience more explicit: The goal of visual prediction is to use priors acquired from both previous experience and the currently unfolding visual stimulus to create a perceived state of the world that matches, as far as possible, the actual state of the world.