Fig 1.
Time course of a trial with the different signals used for each experiment.
A central fixation point is presented for a random duration (ranging from 750 to 1250 ms) then the target appears either to the left or to the right and a brief signal is delivered either before or after target onset (SOA ranging from −240 to +240 ms). The signal was either a beep (exp. A, B & F), a background luminance decrease (exp. C), a pair of isoluminant strips (exp. D) or a tactile vibration (exp. E & F), all lasting about 10 ms. In trials with the baseline conditions, no signal was delivered. The participant had to shift gaze toward the target as fast as possible, regardless of the non-informative signal. The saccade latency is defined as the delay between the target appearance and the eye-movement onset; the saccade duration is the time needed to land on the target.
Fig 2.
Effect of sound beeps on saccade onset. Median latencies (left panel) and median absolute deviations (right panel) averaged across participants for each SOA condition (top) and their corresponding nscores (bottom). Dashed lines show the No beep condition level and grey lines show individual results. Error bars indicate inter-individual SEM. Statistics included a single sample t-test performed on the nscores of the SOA = 0 ms reference condition to highlight the difference with the No beep condition (black arrow), and paired t-tests comparing this reference with each of the other SOA condition (red stars for each SOA above the X-axis). Three stars indicate highly significant differences after Bonferroni correction (p<0.00555) while single stars significant differences without correction (p<0.05). Red arrows indicate the overall range of variations.
Fig 3.
Effect of sound beeps on saccade onset with blocked conditions (left panel). Median latencies averaged across participants for each SOA condition (top) and the corresponding nscores (bottom). Dashed lines show the No beep condition level and grey lines show individual data. Error bars indicate inter-individual SEM. Statistics included a single sample t-test performed on the nscores of the SOA = 0 ms reference condition to highlight the difference with the No beep condition (black arrow), and two paired t-tests comparing this reference with SOA = −60 or +60 ms conditions (red stars above the X-axis). Interleaved vs. blocked conditions (right panel). Comparison between experiment B (red plot) and A (grey plot): median latency differences (top) and nscore differences (bottom) aligned on the No beep condition (dashed lines). Unpaired t-tests performed on the nscore differences between experiments for the 3 SOA conditions showed no significant difference. Three stars indicate highly significant differences after Bonferroni correction (p<0.01666). Red arrows indicate the overall range of variations.
Fig 4.
Effect of background flashes on saccade onset (left panel). Median latencies averaged across participants for each SOA condition (top) and the corresponding nscores (bottom). Dashed lines show the No flash condition level and grey lines show individual results. Error bars indicate inter-individual SEM. Statistics included a single sample t-test performed on the nscores of the SOA = 0 ms reference condition to highlight the difference with the No flash condition (black arrow), and paired t-tests comparing this reference with each of the other SOA condition (red stars for each SOA above the X-axis). Auditory beeps vs. visual background flashes (right panel). Comparison between experiment C (red plot) and A (grey plot): median latency differences (top) and nscore differences (bottom) aligned on the No signal conditions (dashed lines). Statistics were unpaired t-tests performed on the nscore differences between experiments for each of the 9 SOA conditions. Three stars indicate highly significant differences after Bonferroni correction (p<0.00555), single stars and two stars significant differences without correction (p<0.05 and p<0.01 respectively). Red arrows indicate the overall range of variations.
Fig 5.
Effect of strips flashes on saccade onset (left panel). Median latencies averaged across participants for each SOA condition (top) and the corresponding nscores (bottom). Dashed lines show the No flash condition level and grey lines show individual results. Error bars indicate inter-individual SEM. Statistics included a single sample t-test performed on the nscores of the SOA = 0 ms reference condition to highlight the difference with the No flash condition (black arrow), and paired t-tests comparing this reference with each of the other SOA condition (red stars for each SOA above the X-axis). Background flashes vs. strips flashes (right panel). Comparison between experiment D (red plot) and C (grey plot): median latency differences (top) and nscore differences (bottom) aligned on the No flash condition (dashed lines). Statistics were unpaired t-tests performed on the nscore differences between experiments for each of the 9 SOA conditions. Three stars indicate highly significant differences after Bonferroni correction (p<0.00555) while single stars significant differences without correction (p<0.05). Red arrows indicate the overall range of variations.
Fig 6.
Effect of tactile stimulation on saccade onset (left panel). Median latencies averaged across participants for each SOA condition (top) and the corresponding nscores (bottom). Dashed lines show the No touch condition level and grey lines show individual results. Error bars indicate inter-individual SEM. Statistics included a single sample t-test performed on the nscores of the SOA = 0 ms reference condition to highlight the difference with the No touch condition (black arrow), and paired t-tests comparing this reference with each of the other SOA condition (red stars for each SOA above the X-axis). Auditory beeps vs. tactile vibration (right panel). Comparison between experiment E (red plot) and A (grey plot): median latency differences (top) and nscore differences (bottom) aligned on the No signal condition (dashed lines). Statistics were unrelated t-tests performed on the nscore differences between experiments for each of the 9 SOA conditions. Three stars indicate highly significant differences after Bonferroni correction (p<0.00555) while single stars and two stars significant differences without correction (p<0.05 and p<0.01 respectively). Red arrows indicate the overall range of variations.
Fig 7.
Effect of tactile, auditory and audio-tactile stimulation on saccade onset. (Left panel) Median latencies averaged across participants for each SOA condition (top) and their corresponding nscores (bottom). Dashed lines show the No signal condition level. Statistics performed on the nscores for each modality separately included a single sample t-test to compare the SOA = 0 ms reference condition with the No signal condition (colored arrows), and paired t-tests comparing this reference with each SOA condition (colored stars for each SOA above the X-axis). (Top-right) Average signal effect on latency nscores for each modality when SOA = 0 ms. (Bottom-right) Average slopes of the nscores linear regressions computed for each modality within the range [−60 ms, +60 ms]. Grey lines show individual results and error bars indicate inter-individual SEM. Comparisons were done using paired t-tests. Three stars indicate highly significant differences after Bonferroni correction (p<0.01) while single stars significant differences without correction (p<0.05).
Fig 8.
Illustration of how temporal multisensory integration could introduce an illusory gap/overlap for positive/negative SOA. This additional effect would have reduced the extent of the multisensory modulation of saccade latencies by reducing the effect of the perceived target onset time shift.