A unified comparison of stimulus-driven, endogenous mandatory and 'free choice' saccades

Abstract

It has been claimed that saccades arising from the three saccade triggering modes-stimulus-driven, endogenous mandatory and 'free choice'-are driven by distinct mechanisms. We tested this claim by instructing observers to saccade from a white or black fixation disc to a same polarity (white or black) disc flashed for 100 or 200 ms presented either alone (Exo), or together with an opposite (Endo) or same (EndoFC) polarity disc (blocked and mixed sessions). Target(s) and distractor were presented at three inter-stimulus intervals (ISIs) relative to the fixation offset (ISI: -200, 0, +200 ms) and were displayed at random locations within a 4°-to-6° eccentricity range. The statistical analysis showed a global saccade triggering mode effect on saccade reaction times (SRTs) with Endo and EndoFC SRTs longer by about 27 ms than Exo-triggered ones but no effect for the Endo-EndoFC comparison. SRTs depended on both ISI (the "gap-effect"), and target duration. Bimodal best fits of the SRT-distributions were found in 65% of cases with their count not different across the three triggering modes. Percentages of saccades in the 'fast' and 'slow' ranges of bimodal fits did not depend on the triggering modes either. Bimodality tests failed to assert a significant difference between these modes. An analysis of the timing of a putative inhibition by the distractor (Endo) or by the duplicated target (EndoFC) yielded no significant difference between Endo and EndoFC saccades but showed a significant shortening with ISI similar to the SRT shortening suggesting that the distractor-target mutual inhibition is itself inhibited by 'fixation' neurons. While other experimental paradigms may well sustain claims of distinct mechanisms subtending the three saccade triggering modes, as here defined reflexive and voluntary saccades appear to differ primarily in the effectiveness with which inhibitory processes slow down the initial fast rise of the saccade triggering signal.

Figure 1. Spatial and temporal layout of the different experimental conditions.

A. Spatial layout showing fixation (top panel) and target and distractor discs. Fixation in the bottom panel is shown between brackets because, depending on the timing condition, it was or was not present during the display of the target(s)/distractor. The target disc was always of the same color as the fixation disc and was presented by itself (Exo), accompanied by an opposite polarity (Endo) or by a same polarity (EndoFC) disc. Target(s) and distractor where presented at the same eccentricity randomized across trials within the 4-to-6° area indicated by the gray annulus (invisible in the experiments). Target and distractor (Endo) or the two potential targets (EndoFC) were separated by a minimum angle α of 90°. B. Time-sequence of one trial with 100 and 200 ms stimuli (top and bottom panels) for each of the three fixation-target(s)/distractor intervals: gap, simultaneous and overlap conditions (ISI = 200, 0, −200, respectively).

Figure 2. SRTs averaged over observers and conditions to show the effects of Target Duration (A), ISI (B), Saccade Triggering modes (C) and experimental Design (D).

Thin bars are ±1 SE over the 5 observers.

Figure 3. Mean Exo, Endo and EndoFC SRTs under blocked and mixed conditions.

Figure 4. Means across observers (μ1, μ2) of the bimodal SRT-distribution fits (left ordinates, solid bars) and percentages of saccades in the slower SRT-distributions (right ordinates and striped bars) for the blocked and mixed designs (upper and lower panels), the two target durations (left and right panels) and for each of the three triggering modes (Exo: red; Endo, green; EndoFC, blue) as a function of ISI.

Figure 5. Vincentized SRTs (5 observers; bars) and their uni- or bimodal log-normal fits (continuous curves) for all experimental conditions.

As attested by their summary statistics given in Table 1, the uni- vs. bimodal fits show no obvious structured pattern.