Prediction in skilled movements has been studied for more than 50 years (e.g., Gottsdanker Reference Gottsdanker1952). Analysis of eye movements show that baseball and cricket batters and tennis players make an anticipatory saccade to the expected location of the ball and then pursue it foveally for a short period of time (Bahill & LaRitz Reference Bahill and LaRitz1984; Land & Furneaux Reference Land and Furneaux1997; Land & McLeod Reference Land and McLeod2000). In conjunction with physiological evidence discussed below, these data support the view that compensation for neural latencies is based primarily on predictive motor plans (Kerzel & Gegenfurtner Reference Kerzel and Gegenfurtner2003; Öğmen et al. Reference Öğmen, Patel, Bedell and Camuz2004; Patel et al. Reference Patel, Öğmen, Bedell and Sampath2000; Purushothaman et al. Reference Purushothaman, Patel, Bedell and Öğmen1998). Perceptual compensation in early visual areas, as suggested by Nijhawan in the target article, need not occur. Although Nijhawan links neurophysiological findings to “perceptual compensation,” we show here that these links are misconstrued.

As Öğmen et al. point out in their commentary in this BBS issue, Nijhawan fails to distinguish between “prediction” and “compensation.” As a consequence, he mistakes predictive phenomena, such as perisaccadic suppression of retinal motion and remapping of receptive fields (RFs), as evidence for “compensation of neural delays.” For example, the author argues that the perisaccadic remapping of RFs in the lateral intraparietal (LIP) area (Duhamel et al. Reference Duhamel, Colby and Goldberg1992) implies that “sensory processes participate in prediction” (target article, sect. 2, para. 4) and advances a model for perceptual compensation “based on the interaction of signals between the magnocellular and parvocellular systems” (sect. 5.2.1) within area V1 (see Fig. 6 of the target article). However, impending motor action is a prerequisite for the RF remapping reported by Duhamel et al (Reference Duhamel, Colby and Goldberg1992), which indicates that this phenomenon is linked to motor preparation and action and does not imply “compensation of neural delays” at the perceptual level. Indeed, RF remapping was shown not to occur in area V1 (Nakamura & Colby Reference Nakamura and Colby2002). As discussed by Öğmen et al (this issue), the only example in the target article of purely perceptual extrapolation is from Maus and Nijhawan (Reference Maus and Nijhawan2006), which is accounted for by differences in sensitivity.Footnote ¹

Nijhawan attempts to link the flash-lag effect (FLE) to the physiological findings of Berry et al. (Reference Berry, Brivanlou, Jordan and Meister1999), but the FLE increases linearly with the velocity of the moving object (Krekelberg & Lappe Reference Krekelberg and Lappe1999; Nijhawan Reference Nijhawan1994); and Berry et al. showed that the neural response to a moving stimulus shifts increasingly in the opposite direction of motion as the velocity of the moving stimulus is increased. Whether compensation occurs at the perceptual or motor levels can be assessed by considering the dynamic response properties of neurons in areas associated with perception and motor control. Smooth pursuit eye movements exhibit approximately zero phase lag in response to predictable low-frequency target motion (see Carpenter Reference Carpenter1988, pp. 58–63). However, in the absence of timing and direction cues, a pursuit latency of 100–150 msec occurs at motion onset, and this lag gradually approaches zero as the pursuit system evolves from its transient to steady-state mode of operation (Bahill & McDonald Reference Bahill and McDonald1983; Barnes et al. Reference Barnes, Barnes and Chakraborti2000; Fukushima et al. Reference Fukushima, Yamanobe, Shinmei and Fukushima2002). In primates, the middle temporal (MT) area is closely associated with motion perception (see Britten Reference Britten, Chalupa and Werner2004; Britten et al. Reference Britten, Newsome, Shadlen, Celebrini and Movshon1996; Liu & Newsome Reference Liu and Newsome2005; Purushothaman & Bradley Reference Purushothaman and Bradley2005), and the frontal eye field (FEF), a premotor area, is involved in generating eye movements (see Bruce et al. Reference Bruce, Friedman, Kraus, Stanton, Chalupa and Werner2004; Schall Reference Schall, Chalupa and Werner2004). Lesions to MT impair motion perception (Newsome & Paré Reference Newsome and Paré1988), whereas lesions to parts of FEF impair predictive smooth pursuit movements (Keating Reference Keating1991; MacAvoy et al. Reference MacAvoy, Gottlieb and Bruce1991). Even with bilateral FEF lesions, monkeys can still detect target motion and attempt to track it, albeit with very low gain and high phase lag (Keating Reference Keating1991). In addition, the stimulus-driven component of FEF neural responses is modulated significantly by the relevance of the stimulus for impending visuomotor behavior, and modulated little by the physical features of the stimulus (cf. Schall Reference Schall, Chalupa and Werner2004). Finally, electrical stimulation of MT neurons biases perceived movement direction without causing eye movements (Nichols & Newsome Reference Nichols and Newsome2002; Salzman et al. Reference Salzman, Britten and Newsome1990; Reference Salzman, Murasugi, Britten and Newsome1992), whereas stimulation of FEF neurons directly induces eye movements (Gottlieb et al. Reference Gottlieb, Bruce and MacAvoy1993; Reference Gottlieb, MacAvoy and Bruce1994). Thus, the activities of FEF neurons are closely associated with the generation of eye movements and are relatively unrelated to perceptual qualities of the tracking target.

MT neurons show response lags on the order of 100 msec for step-ramp smooth pursuit tasks (Komatsu & Wurtz Reference Komatsu and Wurtz1988; Newsome et al. Reference Newsome, Wurtz and Komatsu1988). Despite complete compensation exhibited by smooth pursuit eye movements for predictable stimuli, these response lags do not diminish for similar stimuli (Lisberger & Movshon Reference Lisberger and Movshon1999), ruling out delay-compensation at or before this cortical site linked to motion perception. In the FEF, neurons active during smooth pursuit (pursuit neurons) veridically track predictable target motion (Gottlieb et al. Reference Gottlieb, Bruce and MacAvoy1993; Reference Gottlieb, MacAvoy and Bruce1994; MacAvoy et al. Reference MacAvoy, Gottlieb and Bruce1991; Tanaka & Fukushima Reference Tanaka and Fukushima1998). However, no FLE occurs during accurate pursuit (Nijhawan Reference Nijhawan2001). Therefore, the linkage of the FLE to lag-compensation requires the analysis of conditions in which no smooth pursuit occurs. In the absence of active pursuit, pursuit neurons in FEF respond very little to target motion, predictable or otherwise (Gottlieb et al. Reference Gottlieb, Bruce and MacAvoy1993; Reference Gottlieb, MacAvoy and Bruce1994; MacAvoy et al. Reference MacAvoy, Gottlieb and Bruce1991; Tanaka & Fukushima Reference Tanaka and Fukushima1998). Pursuit neurons in FEF that do respond to moving targets in the absence of active pursuit show “a clear phase lag due to visual latencies” (Fukushima et al. Reference Fukushima, Yamanobe, Shinmei and Fukushima2002), just like upstream MT and MST neurons. Unlike the motion-sensitive neurons in MT, this lag decreases after approximately one cycle of predictable motion (Fukushima et al. Reference Fukushima, Yamanobe, Shinmei and Fukushima2002).

In summary, the known physiology of motion perception and motor control is in close agreement with our proposal that the locus of compensation for neural time delays is primarily within the motor rather than the perceptual system (Öğmen et al. Reference Öğmen, Patel, Bedell and Camuz2004; Patel et al. Reference Patel, Öğmen, Bedell and Sampath2000; Purushothaman et al. Reference Purushothaman, Patel, Bedell and Öğmen1998).