In this target article, Nijhawan proposes that neural delays in perception must be compensated by visual predictions and internal models. He emphasizes that this compensation is particularly important in interceptive actions and that this issue remains to be addressed. We agree, but would like to point out that there are empirical data and an alternative hypothesis which do not depend on the use of internal models.

In interceptive actions, researchers generally refer to a visuomotor delay (VMD) to define the time period between the pick-up of information and its use in producing an adjustment in movement (e.g., Tresilian Reference Tresilian1993). It has been shown that VMD duration depends on the task (Benguigui et al. Reference Benguigui, Ripoll and Broderick2003). Lee et al. (Reference Lee, Young, Reddish, Lough and Clayton1983) calculated VMDs that ranged from 50 to 135 msec in a ball striking task. Whiting et al. (Reference Whiting, Gill and Stephenson1970) showed, in a ball-catching experiment, that performance did not degrade when the occlusion of the final part of the trajectory was equal or inferior to 100 msec. This period of time was interpreted as a VMD during which no information was used for catching. Bootsma and van Wieringen (Reference Bootsma and van Wieringen1990), in a table-tennis task, observed that the variability was minimal at about 100 msec before contact. This phase of minimal variability was described as the end of the control of the action, reflecting the duration of VMD.

Some researchers have shown that VMD could be longer than 100 msec. McLeod (Reference McLeod1987) reported that expert cricket batsmen, confronted with unexpected changes in the ball trajectory at the bounce, needed at least 190 msec to adapt their swing to the new trajectory. Benguigui et al. (Reference Benguigui, Ripoll and Broderick2003) demonstrated in a task consisting of intercepting a moving stimulus with a thrown projectile, that movements were initiated around 200 msec after a specific source of information had reached a critical value. This delay was interpreted as a VMD between the pick-up of the information and the initiation of movement.

The duration of VMD appears to depend on the use of information. It can be as short as 100 msec when information is used in a continuous mode, but can reach values near 200 msec when information is used in a discrete mode, such as the beginning of the movement or some important correction of that movement.

One can assume that (1) a short VMD and (2) an accurate compensation of it are essential for accurate timing. First, because information is continuously changing during the approach of a moving object, a reduction in VMD allows later pick-up of increasingly accurate information as contact draws near. The later the information pick-up, the more accurate the interceptive action will be (Lobjois et al. Reference Lobjois, Benguigui and Bertsch2006). Le Runigo et al. (Reference Le Runigo, Benguigui and Bardy2005) showed that the time required to produce an adaptation in an interceptive movement after an unexpected change in the moving object's velocity was shorter in expert tennis players (162 msec) than in novices (221 msec). They also showed that the reduction of VMD was highly correlated to timing accuracy, suggesting that a reduced VMD provides the opportunity to improve on-line regulations and to adapt these regulations at later stages before contact.

Second, the movement itself must compensate for VMD under the constraint of an incompressible delay. Moreover, the difficulty in compensating for VMD would increase as VMD increases, as is the case in older adults (Bennett & Castiello Reference Bennett and Castiello1995). Lobjois et al. (Reference Lobjois, Benguigui and Bertsch2006) showed that this increase in VMD explained the lateness of elderly people in a coincidence-timing task. Interestingly, elderly people who had a regular sport activity (e.g., tennis) and who had an increase in VMD that was similar to that of sedentary elderly people were not late in their responses. These results suggest that sport practice also allows a better compensation for the age-related VMD increase.

Although the problem of compensation for VMD has not yet been extensively addressed, some suggestions deserve consideration. When an interceptive movement is very short and controlled in a ballistic mode (movement time: MT≤150 msec; Tresilian et al. Reference Tresilian, Plooy and Carroll2004), the actors have to estimate temporally the ensemble of the visuomotor sequence, including VMD and the MT, and detect the instant at which time-to-collision (TTC) becomes equal to the duration of this sequence (Benguigui et al. Reference Benguigui, Ripoll and Broderick2003).

Analogous mechanisms could be also involved when action must be controlled with continuous regulation. In contrast to Nijhawan's claim (sect. 7.2.1.1), the visual system is not an accurate predictor of moving objects when the last part of the trajectory is occluded beyond a duration of 200 msec (or of a VMD). Numerous studies using prediction motion tasks consisting of estimating the TTC of a moving object after the occlusion of the final part of the trajectory have shown that accuracy and variability in timing responses are dramatically affected by occlusions longer than 200 msec (e.g., Benguigui et al., in press). This means that for MTs above 200 msec, on-line regulation of movement is indispensable for dealing with both unpredictable and predictableFootnote ¹ changes in the object trajectory.

Once again, however, on-line regulations require compensation for VMD. This compensation process can be understood according to the calibration principle laid out by ecological psychologists (Jacobs & Michaels Reference Jacobs and Michaels2006). As motor control is based on laws corresponding to functional relationships between information and movement (Warren Reference Warren, Meijer and Roth1988), the calibration process consists in attuning some parameters of the law of control in order to adapt the movement to the constraints of a specific task. Regarding VMD compensation, laws of control include some parameters that do not correspond to a representation of VMD as is suggested by Nijhawan. Instead, these parameters have to be set according to the task-dependent effect of VMD in the interceptive timing. This process could explain why athletes who master a specific set of skills (exploiting a specific law of control) generally need a few trials of preparation to optimize their efficiency when beginning a new session of practice. This process could also correspond to an issue of learning that remains to be explored.