Introduction 1987 1988 1998 2001 1968 1983 1992 1994 1988 1992 1995 1999 2000 2002 1998 2002 2004 2006 2006 1 Fig. 1 a dark square brighter squares b c d  2003 2006 2003 Materials and methods Participants Twelve subjects took part in this study. Three of them were the authors and the others were colleagues and friends who were unaware of the hypothesis that was being tested. All of them had normal or corrected to normal vision. Two of the participants were left-handed. Each participant made 120 pointing movements in each of 13 different sessions. Six of the participants also made 120 pointing movements in each of 7 additional sessions in which various luminance contrasts were used. Each participant performed the sessions in a different random order. The ethics committee of the Faculty of Human Movement Sciences approved the study. Apparatus and stimuli Participants sat behind an A2-sized graphic tablet (Digitizer II, Wacom Ltd, Tokyo, Japan) and viewed a projection surface via a semitransparent mirror that was placed above this graphic tablet. The images were back-projected from above the projection surface. The distance between the mirror and the projection surface was identical to that between the mirror and the surface of the tablet, so that the projected image appeared to be at the surface of the tablet. Lamps underneath the mirror ensured that the participants could see their hands. The resolution of the projected image was 1,024 by 768 pixels, with 1 pixel corresponding to about 0.5 mm. The position of the pen was determined every 5 ms (200 Hz). 2 2 1 2 2 Fig. 2 left 2 a Red lettering blue lettering b brackets brighter  2 2 2 2 2 2 2 2 During pilot-experiments, some of the participants had the impression (especially in the orientation and the shape conditions) that they were not responding to a difference between the target and the references, but to the change at the new target position. To investigate whether this was really an issue we performed two control conditions in which we masked such effects by also changing aspects of the references when the target changed its position. In a masked orientation condition, the references both changed their orientation by 90° when the target jumped to a new location. In a masked shape condition, the reference squares increased in size at the same moment that the target circle changed position. Procedure 1 1 1 1 Analysis Velocities were calculated for the interval between every two measurements by dividing the displacement of the tip of the pen by the 5 ms between the measurements. The beginning of the movement was defined as the first position after the tangential velocity reached 0.02 m/s. The end of the movement was defined as the first position after the tangential velocity fell below 0.02 m/s. To evaluate the corrections, we only used the lateral component of the velocity (parallel to the displacement of the target). In order to isolate the responses to changes in target position, we first synchronized all the measurements relative to the moment that the target changed—or would have changed—position. We then separately averaged the lateral velocity for each combination of initial and final target position. We defined the lateral velocity in the direction of the position change as being positive; when the position did not change, we defined the direction in which it would have changed as being positive. We characterize the response to a change in target position by the additional lateral velocity: the difference in the lateral component of the velocity between the trials in which the target changed from a certain initial position to a certain final position and the ones in which it remained at the same initial position. To obtain a single response per attribute for each participant we averaged the additional lateral velocity across the six combinations of initial and final target positions. We used these curves to determine the latencies of the responses. 3 t Fig. 3 dotted lines black horizontal line a b  The slope of the line through the 25 and 75% points of the average additional lateral velocity does not only depend on the intensity of the response on individual trials (lower acceleration results in a shallower slope). If the response does not always occur at the same time, averaging will result in an average response that has a lower peak velocity, shallower slope and longer duration than the responses in individual trials. To evaluate the shape of the response curve without the influence of variability in timing, we used a second way of synchronizing trials before determining the additional lateral velocity. To estimate the intensity of individual responses for each condition and participant, we synchronized the lateral velocity curves of all perturbed trials at the peak lateral velocity in the direction of the new target position (irrespective of when the target position changed), and produced an additional lateral velocity curve for each subject by averaging across replications and perturbation directions. We averaged these synchronized additional lateral velocity curves across positions and across participants to investigate whether there are systematic differences between the intensity of the responses for different attributes. We averaged the curves across conditions rather than participants to investigate whether there are systematic differences between the intensity of different participants’ responses. In order to determine whether the response was proportional to the size of the perturbation, we also averaged the curves separately for all pairs of positions in which the target jumped 3 cm, and all pairs of positions in which the target jumped 6 cm. Results Overview of responses 1 1 1 1 3 2 materials and methods 3 3 3 Dealing with conspicuousness 3 4 4 4 4 P Fig. 4 a c 3 Dashed lines d–f Symbols in red symbols in blue g–i  4 4 4 4 2 5 3 2 4 4 Fig. 5 a 4 b c  Different attributes, different latencies? 4 4 2 4 t P 4 t P 4 4 4 4 Subjects and amplitudes 4 5 2 5 2 5 4 4 4 5 5 Discussion 5 An important question in this kind of research is whether the apparent response to a particular attribute could actually be a response to a small difference in another attribute. For instance, in the color condition, participants could respond to small differences in luminance instead of to differences in color between the target and the references, since the luminance of the target and the references was not equated for individual participants. They clearly do not, because if they had done so, the response would be similar to that for one of the small luminance contrast conditions, which is clearly not the case (the slope of the response is much steeper for color). As mentioned earlier, it takes about as long to respond to changes in position defined by target size as to those defined by target luminance. We cannot reject the possibility that the reactions to the difference in surface area between the target and the references in the size condition were actually responses to the average luminance within an area larger than that of the target itself. To be sure that this was not the case we would have had to vary the luminance of the targets and references, which we did not do in the present study. 1999 1983 1994 2000 2005 1997 2000 2003 1983 1987 1988 2003 Although we cannot decide between these two lines of explanation, the two masked conditions indicate what is essential for the very fast responses. Being able to identify the target on the basis of “where” attributes (size, orientation, luminance) is necessary for fast responses. However this is not sufficient, because the latency of responses to the change in location of a target defined by orientation increases by more than 50 ms if the references change their orientation. Masking changes prevents responses to the location of the change. Presumably, the target has to be found (identified) again, which requires ventral processing (according to the first explanation). 2004 3 2004 4 1999