Hand path priming in manual obstacle avoidance: Rapid decay of dorsal stream information

Abstract

The dorsal, action-related, visual stream has been thought to have little or no memory. This hypothesis has seemed credible because functions related to the dorsal stream have been generally unsusceptible to priming from previous experience. Tests of this claim have yielded inconsistent results, however. We argue that these inconsistencies may be due to methodological differences in the time between primes and test stimuli. In this study we sought to clarify the effect of time between primes and test stimuli by having participants complete a visually guided manual obstacle avoidance task with varying times between trials. Consistent with a previous study using this task, we found that hand path curvature depended on the presence or absence of an obstacle in the previous trial. This hand path priming effect decayed quickly as the time between trials increased, and was almost, though not entirely, eliminated when 1000 ms separated successive trials. The results are consistent with the hypothesis that the dorsal stream can be primed but that this priming attenuates rapidly. We suggest that this outcome may indicate that the period over which the dorsal stream retains information may be related to the sequential statistics of action.

Introduction

A fundamental feature of any information processing module is how long it retains information. For example, long-term memory maintains information that can be repeatedly accessed over a prolonged period of time, whereas short-term memory maintains information that can only be accessed for a brief time after the information is stored.

Similar differences in retention time have been proposed for the modules responsible for visual processing. It has long been known (Goodale & Milner, 1992) that separate neural streams process visual information for perceptual identification (the ventral stream) and for the control of action (the dorsal stream). Besides differing in the behavioral functions they perform and their neuroanatomic bases, these two streams also differ in the time span over which they retain information. Ventral stream processing required for picture naming (e.g., van Turennout, Bielamowicz, & Martin, 2003) shows evidence of information retention in the form of priming over a 48-week delay (Cave, 1997). By contrast, dorsal stream processing shows little evidence of long-term retention (Goodale & Westwood, 2004), which may be advantageous when the relationship between the actor and the target of action changes rapidly. Under these conditions, computing movement parameters immediately before movement initiation may be more efficient than relying on stored movement parameters that may no longer be appropriate (Cant, Westwood, Valyear, & Goodale, 2005).

To check whether the dorsal stream fails to retain information, Garofeanu, Kroliczak, Goodale, and Humphrey (2004) asked whether grasping or naming an object would be primed by previous experience with either type of response involving the same object. The authors found that naming was primed by both grasping and by naming primes, consistent with priming within the ventral stream. However, grasping responses were not primed by prior naming or by prior grasping, consistent with a lack of priming within the dorsal stream (see also experiment 3 of Cant et al., 2005).

In contrast to the foregoing studies, other studies have suggested that the dorsal stream may in fact be susceptible to priming. There are reports that the orientation of a preceding visual stimulus can prime a subsequent grasp of a remembered target, a result consistent with dorsal stream priming (Craighero, Bello, Fadiga, & Rizzolatti, 2002; Vogt, Taylor, & Hopkins, 2003). One can argue back, however, that the control of movements to remembered targets may differ from the control of movements to visible targets, with movements made to remembered targets being more reliant on target representations in the ventral stream (Hu & Goodale, 2000; Milner, Paulignan, Dijkerman, Michel, & Jeannorod, 1999). Consistent with this hypothesis, the orientation of a visual stimulus does not prime the grasp of a visible target (Cant et al., 2005) though it can prime the grasp of a remembered target (Craighero et al., 2002, Vogt et al., 2003). Therefore, the priming of grasping movements to remembered targets may occur within the ventral stream.

Grasping has also been shown to be primed by the observation of another's grasp (Castiello, Lusher, Mari, Edwards, & Humphreys, 2002; Edwards, Humphreys, & Castiello, 2003), but again this outcome may have stemmed from non-dorsal-stream areas such as the mirror–neuron system (Rizzolatti, Fogassi, & Gallese, 2001), which occupies the ventral premotor and inferior parietal lobes (an area between the dorsal and ventral streams; Buxbaum, 2001; Pisella, Binkofski, Lasek, Toni, & Rossetti, 2006; Rizzolatti & Matelli, 2003).

Owing to the uncertainty about whether the dorsal stream is primable, Jax and Rosenbaum (2007) conducted an experiment designed to test for priming without the contaminations associated with the studies mentioned above. In Jax and Rosenbaum's study, participants completed a visually guided manual obstacle avoidance task. When obstacle-present and obstacle-absent trials were mixed within a block, the hand paths that participants produced were strongly influenced by the presence or absence of an obstacle on the previous trial. The authors called this effect of the previous trial type hand path priming.

The nature of the hand path priming effect was that when obstacles were likely but did not appear, participants made much more curved hand movements to targets that stood “behind” the likely position of the obstacles than when obstacles never appeared. Of critical importance for the present argument, these needlessly curved hand paths were obtained even where obstacles had never occurred or had occurred many trials ago. This made it unlikely that the hand path priming effect was merely due to memory for obstacles. Similarly, hand paths in obstacle-present trials were less curved if preceded by obstacle-absent trials than if preceded by obstacle-present trials. Jax and Rosenbaum argued that the hand path priming effect was most likely due to priming within the dorsal stream because this stream is known to play a critical role in visually guided direct movements and visually guided obstacle avoidance movements (Culham & Kanwisher, 2001; Goodale & Westwood, 2004; McIntosh, McClements, Dijkerman, Birchall, & Milner, 2004; Schindler et al., 2004).

Why did Jax and Rosenbaum (2007) obtain evidence for dorsal stream priming while other studies (Cant et al., 2005, Garofeanu et al., 2004) failed to do so? One possibility is that the delay between primes and test stimuli varied across these studies. Such delays are known to influence priming (Roediger & McDermott, 1993). In the study of Jax and Rosenbaum, the delay between the primes (completion of a movement) and the test stimuli (presentation of the next target) was relatively short (250 ms). By contrast, the delays were longer in the studies that failed to support dorsal stream priming, ranging from 1250 ms (Cant et al., 2005) to several minutes (Garofeanu et al., 2004). Thus, it is possible that Jax and Rosenbaum obtained the findings they did because their prime-target delays were shorter than in the other studies.

If this interpretation is correct, hand path priming should be reduced or even eliminated when longer times between primes and test stimuli are used. We tested this hypothesis in the present experiment by lengthening the time between successive movements from 250 ms, the value used before, to 600 ms in one condition and 1000 ms in another condition. We predicted that the hand path priming effect would be reduced as the time between movements increased. Such an outcome would be consistent with the view that the dorsal stream is subject to memory decay just like other established neural storage systems.