A body-centred frame of reference drives spatial priming in visual search

Spatial priming in visual search is a well-documented phenomenon. If the target of a visual search is presented at the same location in subsequent trials, the time taken to find the target at this repeated target location is significantly reduced. Previous studies did not determine which spatial reference frame is used to code the location. At least two reference frames can be distinguished: an observer-related frame of reference (egocentric) or a scene-based frame of reference (allocentric). While past studies suggest that an allocentric reference frame is more effective, we found that an egocentric reference frame is at least as effective as an allocentric one (Ball et al. Neuropsychologia 47(6):1585–1591, 2009). Our previous study did not identify which specific egocentric reference frame was used for the priming: participants could have used a retinotopic or a body-centred frame of reference. Here, we disentangled the retinotopic and body-centred reference frames. In the retinotopic condition, the position of the target stimulus, when repeated, changed with the fixation position, whereas in the body-centred condition, the position of the target stimulus remained the same relative to the display, and thus to the body-midline, but was different relative to the fixation position. We used a conjunction search task to assess the generality of our previous findings. We found that participants relied on body-centred information and not retinotopic cues. Thus, we provide further evidence that egocentric information, and specifically body-centred information, can persist for several seconds, and that these effects are not specific to either a feature or a conjunction search paradigm.     

Reduced fields of view are neither necessary nor sufficient for distance underestimation but reduce precision and may cause calibration problems

Watt et al. (Exp Brain Res, 2000, 135:411–416) suggested that a reduced field of view causes objects to appear closer than their physical distance. This suggestion is based on the observation that individuals terminated open-loop prehension prematurely when pretending to grasp a paper rectangle initially viewed through a reduced field of view. We tested Watt et al.s suggestion in an open-loop pointing task. In experiment 1, 21 participants pointed at targets in three locations (20, 30 and 40 cm relative to the starting position) in three viewing conditions (full, 16° and 4° field of view). No difference in accuracy was found between conditions but the reduced field of view led to an increase in end-point variability across trials. We interpret these results as indicating that a reduced field of view decreases precision but does not necessarily affect object localisation. In experiment 2, we asked participants to reach-and-grasp a real object under the same three open-loop viewing conditions but without vision following movement onset. The experimental design ensured that haptic feedback was available, which could be used to calibrate reaching movements. We found that the reduced field of view caused no changes in grasp but we observed changes in the transport kinematics consistent with increased variability in the perceptual estimate of target location. Notably there were no changes in the spatial path (expected from movements to a closer location). In experiment 3, we repeated the Watt et al. design but removed vision and forced participants to rely on memory. In this condition we found the same undershoots as described by Watt et al. We conclude that a reduced field of view is neither necessary nor sufficient for underestimation and suggest that a reduced field of view decreases precision. This can cause participants to undershoot and/or alter the movement kinematics but we argue that such findings cannot be ascribed unambiguously to perceptual underestimation as they may reflect strategic alterations in behaviour.     

Multiple frames of reference for pointing to a remembered target

Pointing with an unseen hand to a visual target that disappears prior to movement requires maintaining a memory representation about the target location. The target location can be transformed either into a hand-centered frame of reference during target presentation and remembered under that form, or remembered in terms of retinal and extra-retinal cues and transformed into a body-centered frame of reference before movement initiation. The main goal of the present study was to investigate whether the target is stored in memory in an eye-centered frame, a hand-centered frame or in both frames of reference concomitantly. The task was to locate, memorize, and point to a target in a dark environment. Hand movement was not visible. During the recall delay, participants were asked to move their hand or their eyes in order to disrupt the memory representation of the target. Movement of the eyes during the recall delay was expected to disrupt an eye-centered memory representation whereas movement of the hand was expected to disrupt a hand-centered memory representation by increasing movement variability to the target. Variability of movement amplitude and direction was examined. Results showed that participants were more variable on the directional component of the movement when required to move their hand during recall delay. On the contrary, moving the eyes caused an increase in variability only in the amplitude component of the pointing movement. Taken together, these results suggest that the direction of the movement is coded and remembered in a frame of reference linked to the arm, whereas the amplitude of the movement is remembered in an eye-centered frame of reference.     

The anti-orienting phenomenon revisited: effects of gaze cues on antisaccade performance

When the eye gaze of a face is congruent with the direction of an upcoming target, saccadic eye movements of the observer towards that target are generated more quickly, in comparison with eye gaze incongruent with the direction of the target. This work examined the conflict in an antisaccade task, when eye gaze points towards the target, but the saccadic eye movement should be triggered in the opposite direction. In a gaze cueing paradigm, a central face provided an attentional gaze cue towards the target or away from the target. Participants (N = 38) generated pro- and antisaccades to peripheral targets that were congruent or incongruent with the previous gaze cue. Paradoxically, facilitatory effects of a gaze cue towards the target were observed for both the pro- and antisaccade tasks. The results are consistent with the idea that eye gaze cues are processed in the task set that is compatible with the saccade programme. Thus, in an antisaccade paradigm, participants may anti-orient with respect to the gaze cue, resulting in faster saccades on trials when the gaze cue is towards the target. The results resemble a previous observation by Fischer and Weber (Exp Brain Res 109:507–512, 1996) using low-level peripheral cues. The current study extends this finding to include central socially communicative cues.     

The effects of landmarks on the performance of delayed and real-time pointing movements

Converging lines of evidence suggest that the presence of non-target landmarks affects the performance of delayed target-directed movements (e.g., Diedrichsen et al. 2004; Sheth and Shimojo 2004). In the present experiment, we examined the effects of non-target landmarks on the accuracy and precision of delayed and immediate target-directed pointing movements. In our experiment, the landmarks were present just prior to and during the presentation of the target; they were never present during the execution of the movement. Absolute errors were significantly reduced when the landmarks were available during target presentation for both delayed and immediate action conditions. In contrast, the presence of landmarks improved the precision of delayed but not immediate movements (as indexed by the variable error). The locus of this “landmark benefit” appears to be in the encoding of target position since landmarks were never available after target offset. We suggest that, when available, information provided by landmarks is used to improve the accuracy of the estimation of target location. Since the targets were presented for only 100 ms, it is apparent that the spatial information available from landmarks can be quite rapidly used to estimate target position. Further, with respect to the precision of movements, we suggest that the presence of landmarks serves to improve the stability of the estimation of target position. This particular reliance on landmark information becomes more critical as the movement is delayed.     

Reach adaptation to explicit vs. implicit target error

The adaptation of reaching movements has typically been investigated by either distorting visual feedback of the reaching limb or by distorting the forces acting upon the reaching limb. Here, we investigate reach adaptation when error is created by systematically perturbing the target of the reach rather than the limb itself (Magescas and Prablanc in J Cogn Neurosci 18: 75–83, 2006). Specifically, we investigate how adaptation is affected by (1) the timing of the perturbation with respect to the movement of the eye and the hand and (2) participant awareness of the perturbation. In Experiment 1, participants looked and pointed to a target that disappeared either at the onset of their eye movement or shortly after their eye movement and then reappeared, displaced to the right, at the completion of the reach. In Experiment 2, we made the target displacement more explicit by leaving the target at its initial location until the end of the reach, at which point it was displaced to the right. In Experiment 3, we extinguished the target at the onset of the eye movement but also informed participants about the presence and magnitude of the perturbation. In the no-feedback post-test phase, participants for whom the target disappeared during the reach demonstrated much stronger aftereffects of the perturbation, misreaching to the right, whereas participants for whom the target stayed on until reach completion demonstrated rapid extinction of rightward misreaching. Furthermore, participants who were informed about the target perturbation exhibited faster de-adaptation than those who were not. Our results suggest that adaptation to a target displacement is contingent on the explicitness of the target perturbation, whether this is achieved by manipulating stimulus timing or instruction.     

Returning home: location memory versus posture memory in object manipulation

Previous studies of object manipulation have suggested that when participants return an object to the place from which they just carried it, they tend to grasp the object for the target-back-to-home trips close to where they just grasped it for the home-to-target trips [Exp Brain Res 157(4):486–495, 2004; Psychon Bull Rev, 2006]. What was unclear from these previous studies was whether participants recalled postures or locations. According to the posture hypothesis, they remembered what body positions they adopted when they last held the object. According to the location hypothesis, they remembered where they held the object and then took hold of it there or nearby again. To distinguish between these possibilities, we had participants mount or dismount a platform after home-to-target moves and before target-back-to-home moves. In the control condition, they did not change their vertical position relative to the shelf containing the home and target platforms (they merely stepped sideways). We found that participants grasped the object at nearly the same place along its length as they had before, even if this meant adopting very different postures than before. This outcome is consistent with the location-recall account and is inconsistent with the posture-recall account. The implications for motor planning are discussed.     

Startle response is dishabituated during a reaction time task

Recent experiments pairing a startling stimulus with a simple reaction time (RT) task have shown that when participants are startled, a prepared movement may be triggered earlier in comparison to voluntary initiation (Carlsen et al. 2003, in press; Siegmund et al. 2001; Valls-Solé et al. 1999). The use of this paradigm in experiments may provide new insights into processes that control rapid voluntary actions. However, because the startle response habituates with repeated exposure to the startling stimulus, its use in experiments may be limited. Previously Brown et al. (1991) and later Siegmund et al. (2001) noted that individuals habituate to a startling stimulus at different rates depending on the required activity level of the participant in the task. The present experiment was designed to determine the rate at which participants habituate to a startle during the completion of a RT task. Participants completed 100 trials in which an active wrist extension to a target was performed as fast as possible following an auditory tone. An unexpected 124 dB auditory startle stimulus accompanied the imperative stimulus in 20 of these trials. For the duration of the experiment, startle response electromyographic (EMG) activity continued to be produced in the sternocleidomastoid muscle (SCM) indicating that habituation was not complete after 20 startle trials. Furthermore RT in the startle condition was significantly shorter than control RT. However, findings indicate that when a measurable EMG burst in the SCM was present, RT was significantly shorter than when no SCM burst was present.Anthony N. Carlsen, Romeo Chua, J. Timothy Inglis, David J. Sanderson, and Ian M. Franks, School of Human Kinetics, University of British Columbia, Vancouver, Canada, and the International Collaboration on Repair Discoveries (ICORD). This research was supported by a grant from the Natural Sciences and Engineering Research Council of Canada awarded to Ian M Franks.     

Effects of altering initial position on movement direction and extent

The purpose of this study was to examine the relative influence of initial hand location on the direction and extent of planar reaching movements. Subjects performed a horizontal-plane reaching task with the dominant arm supported above a table top by a frictionless air-jet system. A start circle and a target were reflected from a horizontal projection screen onto a horizontally positioned mirror, which blocked the subject's view of the arm. A cursor, representing either actual or virtual finger location, was only displayed between each trial to allow subjects to position the cursor in the start circle. Prior to occasional "probe trials," we changed the start location of the finger relative to the cursor. Subjects reported being unaware of the discrepancy between cursor and finger. Our results indicate that regardless of initial hand location, subjects did not alter the direction of movement. However, movement distance was systematically adjusted in accord with the baseline target position. Thus when the hand start position was perpendicularly displaced relative to the target direction, neither the direction nor the extent of movement varied relative to that of baseline. However, when the hand was displaced along the target direction, either anterior or posterior, movements were made in the same direction as baseline trials but were shortened or lengthened, respectively. This effect was asymmetrical such that movements from anterior displaced positions showed greater distance adjustment than those from posterior displaced positions. Inverse dynamic analysis revealed substantial changes in elbow and shoulder muscle torque strategies for both right/left and anterior/posterior pairs of displacements. In the case of right/left displacements, such changes in muscle torque compensated changes in limb configuration such that movements were made in the same direction and to the same extent as baseline trials. Our results support the hypothesis that movement direction is specified relative to an origin at the current location of the hand. Movement extent, on the other hand, appears to be affected by the workspace learned during baseline movement experience.     

Inhibition of return in cue–target and target–target tasks

Inhibition of return (IOR), the term given for the slowing of a response to a target that appeared at the same location as a previously presented stimulus, has been studied with both target–target (TT; participants respond to each successive event) and cue–target (CT; participants only respond to the second of two events) tasks. Although both tasks have been used to examine the processes and characteristics of IOR, few studies have been conducted to understand if there are any differences in the processes that underlie the IOR that results from ignoring (CT paradigm) or responding to (TT paradigm) the first stimulus. The purpose of the present study was to examine the notion that IOR found in TT tasks represents “true” IOR whereas IOR found in CT tasks consist of both “true” IOR and response inhibition (Coward et al. in Exp Brain Res 155:124–128, 2004). Consistent with the pattern of effects found by Coward et al. (Exp Brain Res 155:124–128, 2004), IOR was larger in the CT task than in the TT task when a single detection response was required (Experiment 1). However, when participants completed one of two spatially-directed responses (rapid aiming movement to the location of the target stimulus), IOR effects from the CT and TT tasks were equal in magnitude (Experiment 2). Rather than CT tasks having an additional response inhibition component, these results suggest that TT tasks may show less of an inhibitory effect because of a facilitatory response repetition effect.     

Assessing the audiotactile Colavita effect in near and rear space

The Colavita effect occurs when participants performing a speeded detection/discrimination task preferentially report the visual component of pairs of audiovisual or visuotactile stimuli. To date, however, researchers have failed to demonstrate an analogous effect for audiotactile stimuli (Hecht and Reiner in Exp Brain Res 193:307–314, 2009). Here, we investigate whether an audiotactile Colavita effect can be demonstrated by manipulating either the physical features of the auditory stimuli presented in frontal (Experiment 1) or rear space (Experiment 3), or the relative and absolute position of auditory and tactile stimuli in frontal (Experiment 2) or rear space (Experiment 3). The participants showed no evidence of responding preferentially to one of the sensory components of the bimodal stimuli when they were presented from a single location in frontal space (Experiment 1). However, a significant audiotactile Colavita effect was demonstrated in Experiments 2 and 3, with participants preferentially reporting the auditory (rather than tactile) stimulus on the bimodal target trials. In Experiment 3, an audiotactile Colavita effect was reported for auditory white noise bursts but not for pure tones and selectively for those stimuli presented from the same (rather than from the opposite) side. Taken together, these results therefore suggest that when a tactile and an auditory stimulus are presented from a single frontal location, participants do not preferentially report one of the two sensory components (Experiment 1). In contrast, when the stimuli are presented from different locations, people preferentially report the auditory component, especially when they are spatially coincident (Experiments 2 and 3). Moreover, for stimuli presented from rear space, the Colavita effect was only observed for auditory stimuli consisting of white noise bursts (but not for pure tones), suggesting that this kind of stimuli are more likely to be bound together with somatosensory stimuli in rear space.     

Effects of altered transport paths and intermediate movement goals on human grasp kinematics

It has been observed that grip opening is delayed when participants are asked to execute complex grasping movements, such as passing over an obstacle or a via-position (Haggard and Wing 1998; Alberts et al. 2002). This finding was proposed to indicate a shift toward sequential performance, meaning that complex movements are carried out in independent motor steps. In our experiments we investigated which aspects of a grasping task determine whether a movement is executed holistically or sequentially. Therefore, participants had to perform different types of curved movements in order to reach and grasp a target object. When only the complexity of the transport paths was varied, no indication of sequential movement execution was found. However, when participants additionally had to either stop at, or pass over a certain via-position the pre-shaping pattern changed considerably indicating a movement segmentation effect. This effect became stronger with increasing difficulty of the sub-task, suggesting that attentional factors are involved.     

Limb position drift: implications for control of posture and movement

In the absence of visual feedback, subject reports of hand location tend to drift over time. Such drift has been attributed to a gradual reduction in the usefulness of proprioception to signal limb position. If this account is correct, drift should degrade the accuracy of movement distance and direction over a series of movements made without visual feedback. To test this hypothesis, we asked participants to perform six series of 75 repetitive movements from a visible start location to a visible target, in time with a regular, audible tone. Fingertip position feedback was given by a cursor during the first five trials in the series. Feedback was then removed, and participants were to continue on pace for the next 70 trials. Movements were made in two directions (30 degrees and 120 degrees ) from each of three start locations (initial shoulder angles of 30 degrees, 40 degrees, 50 degrees, and initial elbow angles of 90 degrees ). Over the 70 trials, the start location of each movement drifted, on average, 8 cm away from the initial start location. This drift varied systematically with movement direction, indicating that drift is related to movement production. However, despite these dramatic changes in hand position and joint configuration, movement distance and direction remained relatively constant. Inverse dynamics analysis revealed that movement preservation was accompanied by substantial modification of joint muscle torque. These results suggest that proprioception continues to be a reliable source of limb position information after prolonged time without vision, but that this information is used differently for maintaining limb position and for specifying movement trajectory.     

Visuomotor memory is independent of conscious awareness of target features

A recent study by our group showed that the scaling of reach trajectories to target size is independent of conscious visual awareness of that intrinsic target property (Binsted et al. in Proc Natl Acad Sci USA 104:12669–12672, 2007). The present investigation sought to extend previous work and determine whether unconscious target information represents a temporally durable or evanescent visuomotor characteristic. To accomplish that objective, we employed Di Lollo et al’s (J Exp Psychol Gen 129:481–507, 2000) object substitution masking paradigm and asked participants to complete verbal reports and reaching responses to different sized (1.5, 2.5, 3.5, 4.5, 5.5 cm) targets under masked and non-masked target conditions. To determine whether visuomotor networks retain unconscious target information, reaching trials were cued concurrent with target presentation or 1,000 or 2,000 ms after target presentation. For the perceptual trials, participants readily identified the size of non-masked trials but demonstrated only chance success identifying target size during masked trials. Interestingly, however, reaches directed to non-masked and masked targets exhibited comparable and robust scaling with target size; that is, lawful speed-accuracy relations related to movement planning and execution times were observed regardless of whether participants were aware (i.e., non-masked trials) or unaware (i.e., masked trials) of target size. What is more, the length of the visual delay period used here did not differentially influence the scaling of reach trajectories. These results indicate that a conscious visual percept is not necessary to support motor output and that unconscious visual information persists in visuomotor networks to support the kinematic parameterization of action.

The timing of color and location processing in the motor context

In this study, the use of color and location as stimulus attributes manipulated during a simple action was aimed at comparing how dorsal (location) and ventral (color) features are integrated in action and the timing of their processing. Eighteen subjects were presented with a green dot on a computer screen, which they were required to point at and touch. In 20% of the trials, the location or the color of the target was altered at the onset of movement to this stimulus, requiring the participant to modify the initially programmed response according to specific motor instructions. In the ’location-go’ group, the target changed in location and participants were instructed to reach the displaced stimulus by correcting their ongoing movement. In the ’location-stop’ and ’color-stop’ groups, subjects were instructed to interrupt their movement when the target changed location or color, respectively. Results showed that the latency of the first responses to the perturbation clearly depended on the stimulus attribute and not on the motor instruction tested: the response to color change was obtained about 80 ms later than both conditions involving location change. It is concluded that: (1) color processing is slower than location processing, and (2) the first reactions to the location change occur after the same delay irrespective of the response required from the subject. Received: 1 September 1997 / Accepted: 5 February 1998     

Seeing all the obstacles in your way: the effect of visual feedback and visual feedback schedule on obstacle avoidance while reaching

Human reaching behaviour displays sophisticated obstacle avoidance. Recently, we demonstrated that the obstacle avoidance system in normal participants is sensitive to both the position and size of obstacles (Chapman and Goodale in Exp Brain Res 191:83–97, 2008). A limitation in this previous study was that reaches were performed without visual feedback, and were not made to a specific target (i.e. the target was a long strip instead of a point). Many studies have shown that both the introduction of visual feedback and the order in which the feedback is presented (visual feedback schedule) significantly alter performance in simple visuomotor tasks (Zelaznik et al. in J Mot Behav 15:217–236, 1983). Thus, the present study was designed to compare obstacle avoidance when reaches were made to a discrete target with vision (V) and with no vision (NV) under different three visual feedback schedules (blocked, random, and alternating). Twenty-four right-handed participants performed reaches in the presence of one, two, or no obstacles placed mid-reach. In addition to replicating previous work with reaching without vision, we showed that robust avoidance behaviour occurred when reaches were made to a specific target, when reaching with only one object present, and, critically, when vision of the hand was available during the reach. Moreover, the visual feedback schedule also had a significant effect on several kinematic measures—but only on the NV trials. That is, regardless of its predictability or recent availability, vision was used in the same way for all reaches. In contrast, performance on blocked-NV trials was markedly different from performance on NV trials presented under random or alternating schedules. In addition to extending our understanding of obstacle avoidance during reaching, our results suggest that, in a complex and more natural reach-to-point task, the human visuomotor system is optimized to use visual feedback.     

A distance effect in a manual aiming task to remembered targets: a test of three hypotheses

It has been noted that manual aiming error and variability when pointing to remembered targets increase as a function of target eccentricity. In the present study we evaluated which one of three hypotheses (target localization, motor, or movement duration) best explains this 'distance effect'. In experiment 1, older and younger participants aimed with their unseen hand at the remembered location of targets distributed between 129 and 309 mm from the starting base. Target presentation time was of either 50 or 500 ms and aiming movements could be initiated following either a 100- or a 10,000-ms recall delay. Participants had either no constraints concerning movement time or were asked to reach the near target in a longer movement time than the farther targets. The results revealed a significant distance effect when no time constraints were imposed but showed a significantly reversed distance effect when the instructions were to reach the near targets in a longer movement time than the far targets. The same results were obtained regardless of target presentation time, recall delay, or age of the participants. These results supported a movement duration interpretation of the distance effect. In experiment 2, a distance effect was replicated when pointing with one's unseen hand toward a remembered target but did not take place when pointing to visible targets. Taken together these results suggest that prolonged movement execution interferes with the stored egocentric target representation.     

Bimanual reaches with symbolic cues exhibit errors in target selection

We examined the movement trajectories of symmetric and asymmetric bimanual reaches to targets specified by direct spatial cues and by indirect symbolic cues. Symbolically cued asymmetric reaches have been shown to exhibit longer reaction times compared with symmetric reaches, whereas no such reaction time cost is observed when targets are spatially cued—a pattern thought to implicate increased demands on response selection (Diedrichsen et al. in Psychol Sci 12(6):493–498, 2001). As symbolically cued reaches impose greater demands on cognitive visuomotor translation than spatially cued reaches (Diedrichsen et al. in Cereb Cortex 16(12):1729–1738, 2006), we asked whether bimanual movements exhibit more spatial coupling with symbolic cues than with spatial cues. Participants made bimanual symmetric and asymmetric reaches to short- and long-distance targets cued either symbolically or spatially. We replicated the reaction time cost for symbolically cued asymmetric movements. A subset of these asymmetric reaches also showed large trajectory modulations. It appeared that this subset had been incorrectly prepared and the movements required of the left and right arms had been switched. No such errors in target selection were observed when targets were spatially cued. In contrast to the reaction time cost and errors in selection for symbolically cued movements, we observed little evidence of increased spatial coupling with symbolic cues when movements were initiated towards the correct targets. We conclude that cognitive visuomotor translation demands during response selection increases bimanual coupling at the level of response selection (reaction time cost, errors in target selection) but not at the level of movement execution (spatial coupling).     

Differential contributions of vision and proprioception to movement accuracy

We examined the relative roles of visual and proprioceptive information about initial hand position on movement accuracy. A virtual reality environment was employed to dissociate visual information about hand position from the actual hand position. Previous studies examining the effects of such dissociations on perception of hand location have indicated a bias toward the visually displayed position. However, an earlier study, which employed optical prisms to dissociate visual and proprioceptive information prior to targeted movements, suggested a bias in movement direction toward that defined by the actual hand position. This implies that visual and proprioceptive information about hand position may be differentially employed for perceptual judgments and movement planning, respectively. We now employ a virtual reality environment to systematically manipulate the visual display of the hand start position from the actual hand position during movements made to a variety of directions. We asked whether subjects would adjust their movements in accord with the virtual or the actual hand location. Subjects performed a series of baseline movements toward one of three targets in each of three blocks of trials. Interspersed among these trials were "probe" trials in which the cursor location, but not the hand location, was displaced relative to the baseline start position. In all cases, cursor feedback was blanked at movement onset. Our findings indicated that subjects systematically adjusted the direction of movement in accord with the virtual, not the actual, start location of the hand. These findings support the hypothesis that visual information about hand position predominates in specifying movement direction.     

The preparation and control of reversal movements as a single unit of action

Previous research has demonstrated that movement time and kinematic properties of limb trajectories to the first target of a two-target reversal movement differ to that of single-target responses. In the present study we investigated whether two-target reversal movements are organized as a single unit of action or two separate components by perturbing the number of targets prior to and during movement execution. In one experiment, participants performed single-target movements and on one-third of the trials a second target was presented either at target presentation, movement onset or peak velocity. On those trials in which a second target was presented, participants were required to complete their movement to the first target and then move to the second target. In a second experiment, the reverse was the case with participants performing two-target movements that changed to single-target movement on one-third of the trials. A two-target movement time advantage was observed only when the required response was specified prior to movement initiation. Also, participants failed to prevent movement towards the second target when the requirements of the task changed from a two-target to single-target response at movement onset or later. These results indicate that two-target reversal movements were organized as a single unit of action prior to response initiation.