External location of touch is constructed post-hoc based on limb choice

Abstract

When humans indicate on which hand a tactile stimulus occurred, they often err when their hands are crossed. This finding seemingly supports the view that the automatically determined touch location in external space affects limb assignment: the crossed right hand is localized in left space, and this conflict presumably provokes hand assignment errors. Here, participants judged on which hand the first of two stimuli, presented during a bimanual movement, had occurred, and then indicated its external location by a reach-to-point movement. When participants incorrectly chose the hand stimulated second, they pointed to where that hand had been at the correct, first time point, though no stimulus had occurred at that location. This behavior suggests that stimulus localization depended on hand assignment, not vice versa. It is, thus, incompatible with the notion of automatic computation of external stimulus location upon occurrence. Instead, humans construct external touch location post-hoc and on demand.

Keywords: human; localization; movement; neuroscience; reaching; remapping; sensorimotor; tactile.

Figure 1.. Experimental conditions of Experiment 1 and predictions of the tested tactile localization hypotheses.

(A-B) Experimental procedure. (A) The arms moved from an uncrossed or crossed start posture to an uncrossed or crossed arm end posture. (B) Representative examples of TOJ trials showing the bimanual movement (gray, left hand; yellow, right hand) for the four combinations of uncrossed and crossed start and end postures, as well as the reach-to-point movement of the hand at which the first tactile stimulus was reported. (C) Illustation of a correct TOJ trial: the stimulus is assigned to the correct hand, which points to the correct location. Gray (yellow) traces illustrate the left (right) hand’s movement toward the body, here during a trial from an uncrossed start to an uncrossed end posture. The blue arrow indicates the movement of the correctly assigned hand toward the location of the first stimulus (cross). (D-F) Illustration of the three hypotheses that may account for TOJ errors. The red arrows indicate the movement of the incorrectly chosen hand. (D) Space-to-limb reconstruction hypothesis: participants point with the incorrect hand at the external location of the first stimulus. (E) Stimulus switch hypothesis: participants point with the incorrect hand at the external location of the second stimulus. (F) Time reconstruction hypothesis: participants point with the incorrect hand at the location at which that hand was at the time of the first stimulus.

Figure 2.. TOJ task performance in Experiment 1.

Proportion of correct hand assignment across movement conditions (uncrossed-uncrossed, uncrossed-crossed, crossed-uncrossed, crossed-crossed) in the four phases of the bimanual movement (before movement, during start posture, during end posture, after movement). For conditions without a postural change (i.e., uncrossed-uncrossed, crossed-crossed), trials were assigned as ‘during start posture’ if the first stimulus occurred during the first temporal half of the movement and they were assigned as ‘during end posture’ if the first stimulus occurred during the second temporal half of the movement. For conditions with a postural change (i.e., uncrossed-crossed, crossed-uncrossed), trials were assigned as ‘during start posture’ if the first stimulus occurred before the postural change and as ‘during end posture’ if the first stimulus occurred after the postural change. Error bars denote 2 s.e. from the mean; asymmetry is due to nonlinear conversion from the GLMM’s logit scale to percentage correct. Large symbols are group means, small symbols are individual participants’ performance.

Figure 3.. Localization errors systematically vary with the progression of the movement.

(A) Distance of reported hand position from hand position at movement start (‘perceived distance’; calculated as difference in the direction of the trajectory of the reporting hand) plotted against the distance of true hand position at the time of tactile stimulation from movement start (‘true distance’) of all correct TOJ trials of a single participant in Experiment 1. Each dot represents a single trial. The solid line represents veridical localization. (B) Mean localization error (teal line) of all correct TOJ trials of the same participant as in A. Each dot represents the localization error of a single trial, that is the difference between indicated hand position at the end of the trial and true hand position at the time of tactile stimulation. Blue shading represents the average movement time, with 0 ms = movement start. Note, that the localization error is positive at the beginning of the movement, indicating error in the direction of the movement. Localization error is negative toward the end of the movement, indicating error against the direction of movement. (C-D) The localization error pattern of the correct trials in B was evident across all participants for both the 2 stimulus experimental task (C) and for the 1 stimulus control task (D) and for all posture combinations. Traces reflect the group mean, shaded areas around the traces reflect s.e.m. The shaded regions in the background indicate the average movement duration, which differed slightly between conditions (see in Supplementary file 1).

Figure 4.. Localization responses of Experiment 1 are at odds with the space-to-limb reconstruction hypothesis.

Average reach trajectories (solid lines with finger position at movement onset indicated as circles, and at movement offet as squares) and localization responses (i.e., finger positions in the horizontal plane at the end of the reach-to-point movement indicating the location where the participant perceived the first stimulus) for the different movement conditions. Data are from the same participant as in Figure 3AB. Ellipses represent 95% of the variability of localization responses and show large overlap for correct and incorrect TOJ trials. The space-to-limb reconstruction hypothesis would predict that, during error trials, participants point with the incorrectly assigned hand to the location of the correct stimulus; thus, if this hypothesis were correct, orange ellipses should overlay with light gray ellipses, and dark gray ellipses should overlay with yellow ellipses.

Figure 5.. Localization curves, averaged across participants, for each of the four posture conditions in Experiment 1.

Curves of incorrect TOJ trials (red) show a similar pattern as the localization curves of the correct TOJ trials at time 1 (dark blue), but not as the localization curves of the correct TOJ trials at time 2 (light blue). Traces reflect the mean localization error, shaded areas around the traces reflect s.e.m. across participants. The shaded regions in the background represent the average movement time.

Figure 6.. Time shift of stimulus localization error in TOJ error trials relative to time 1 (left panel) and time 2 (middle panel) for the four posture conditions in Experiment 1.

The temporal shift of the localization error curve was significantly different from zero when calculated relative to time 2, but not relative to time 1; this result is consistent with the time reconstruction hypothesis, but not with the stimulus switch hypothesis. This pattern was similar across all participants as demonstrated by the differences in time shift between time 2 and time 1 (right panel). Data are visualized with raincloud plots (Allen et al., 2019) that display probability density estimates, condition averages (large symbols), and individual participants (small symbols). Error bars denote 95% confidence intervals.

Figure 7.. Localization curves, averaged across participants and posture, for each of the four SOAs in Experiment 2.

Curves of incorrect TOJ trials (red) show a similar pattern as the localization curves of the correct TOJ trials at time 1 (dark blue), but not as the localization curves of the correct TOJ trials at time 2 (light blue). This pattern was highly similar across all participants and also when calculated separately for each posture condition (see Supplementary Information). Traces reflect the mean, shaded areas around the traces reflect s.e.m. The shaded regions in the background represent the average movement time.

Figure 7—figure supplement 1.. TOJ performance of Experiment 2.

Proportion of correct hand assignment across movement conditions (uncrossed-uncrossed, crossed-crossed) and SOA (60, 85, 110, 135 ms) in Experiment 2. Error bars denote 2 s.e. from the mean; asymmetry is due to nonlinear conversion from the GLMM’s logit scale to percentage correct. Large symbols are group means, small symbols are individual participants’ performance. TOJ performance in Experiment 2 was modulated by hand posture and SOA. A GLMM with factors posture (uncrossed-uncrossed, crossed-crossed) and SOA (60, 85, 110, 135 ms) revealed significant main effects of posture (χ²(8,9)=586.94, p<0.001) and SOA (χ²(6,9)=218.00, p<0.001), and a significant interaction (χ²(6,9)=66.63, p<0.001). Post hoc analysis of the interaction (Bonferroni corrected, see in Supplementary file 1) showed that TOJ performance was better when the arms were in an uncrossed compared to a crossed posture at all SOAs. Furthermore, performance increased with SOA duration for the uncrossed posture but was relatively similar across all SOAs for the crossed posture.

Figure 7—figure supplement 2.. Localization performance in the uncrossed-uncrossed condition of Experiment 2.

Localization error curves of the uncrossed-uncrossed posture condition, averaged across participants, for each of the four SOAs in Experiment 2. Curves of incorrect TOJ trials (red) show a similar pattern as the localization curves of the correct TOJ trials at time 1 (dark blue), but not as the localization curves of the correct TOJ trials at time 2 (light blue). Traces reflect the mean, shaded areas around the traces reflect s.e.m. The shaded regions in the background represent the average movement time.

Figure 7—figure supplement 3.. Localization performance in the crossed-crossed condition of Experiment 2.

Localization error curves of the crossed-crossed posture condition, averaged across participants, for each of the four SOAs in Experiment 2. Curves of incorrect TOJ trials (red) show a similar pattern as the localization curves of the correct TOJ trials at time 1 (dark blue), but not as the localization curves of the correct TOJ trials at time 2 (light blue). Traces reflect the mean, shaded areas around the traces reflect s.e.m. The shaded regions in the background represent the average movement time.

Figure 7—figure supplement 4.. Single participant example of localization performance in Experiment 2.

Localization error curves of a representative participant (#07), averaged across posture, for each of the four SOAs in Experiment 2. Curves of incorrect TOJ trials (red) show a similar pattern as the localization curves of the correct TOJ trials at time 1 (dark blue), but not as the localization curves of the correct TOJ trials at time 2 (light blue).

Figure 8.. Time shift of stimulus localization error in incorrect TOJ trials relative to time 1 (left panel) and time 2 (middle panel) for the four SOAs in Experiment 2.

For all SOAs, the temporal shift relative to time 1 was not significantly different from 0. In contrast, the time shift was significantly different from zero for all SOAs when calculated relative to time 2, and it was numerically similar to the respective SOA. These results are consistent with the time reconstruction hypothesis, but not with the stimulus switch hypothesis. This pattern was similar across all participants as demonstrated by the differences in time shift between time 2 and time 1 (right panel). Data are visualized with raincloud plots (Allen et al., 2019) displaying probability density estimates, condition averages (large symbols), and individual participants (small symbols). Error bars denote 95% confidence intervals.