Temporal order judgments and presaccadic shifts of attention: What can prior entry teach us about the premotor theory?

A temporal order judgment (TOJ) 2-alternative forced choice design was used to examine presaccadic shifts of attention. Prior work on the premotor theory of attention (PTA) has predominantly focused on single-target discrimination tasks as a tool to measure accuracy and shifts of attention. It is important to demonstrate that the PTA is effective across attentional tasks that have been shown to be reliable in other contexts. Therefore, it was decided to use a perceptual task that probes multiple locations simultaneously and can equally be used to examine spatial spread of attention in more detail. In typical TOJ studies, prior entry is the metric used to measure an attentional effect. Prior entry is the biasing of temporal perception toward an attentionally cued location. This generally manifests as observers processing events at the cued location more rapidly, altering their perspective of temporal order. Participants were required to prepare saccades toward one of four targets, two of which would light up either synchronously or sequentially after a GO signal but before saccadic execution. Results demonstrated that in conditions with critical stimulus onset asynchronies, saccade preparation had a significant effect on performance. Prior entry effects were observed at saccade congruent locations with probes at these locations being typically perceived earlier than probes presented at a neutral location. These effects were not observed in control trials without a saccade. A further spatial effect was demonstrated for the attentional modulation, suggesting that this effect is restricted predominantly to horizontal configurations. Overall, results demonstrated that presaccadic attention is effective at eliciting a prior entry effect in TOJ designs and that such effects are more pronounced when the probes are distributed across the two lateral hemifields.


Figure 1 supplementary material. Delta between successive screen flips:
Results of adaptive sync mode in PsychToolBox. It submits OpenGL bufferswaps/flips of varying delay between successive flips and then measures and plots how well the hardware can follow the requested timing. Figures on the left side represent results with G-Sync technology switched off, and figures on the right show the results with G-Sync turned on. The X axis for all displays refers to the number of frames. A represents the actual measured delay (red) between successive flips against the requested optimal duration (purple). B shows the difference (green) and median error (blue). C shows a histogram of the difference between requested and actual frame duration in msecs.

Procedure
Participants were required to respond to different combinations of probes in a particular manner.
There were three possible probe distributions: horizontal, vertical, and diagonal. See Figure 2 below for more specific details.

Figure 2 supplementary material.
An image detailing the way that key mapping and directionality worked in the study. A demonstrates a typical horizontal trial, as can be seen a horizontal trial was when the participant perceived the two probes crossing the vertical meridian but not the horizontal. This required participants to respond with left and right arrow keys. B shows a typical vertical trial. Here the participant would respond with up and down arrow keys whenever the two probes crossed the horizontal meridian but not the vertial. Finally C, this demonstrates the diagonal condition. This occurred when participants were presented with probes that crossed both the horizontal and vertical meridian and required a key response using left and right arrow keys. Participants were simply taught that any time the two probes crossed the vertical meridian (appeared left and right of fixation) they would respond with the left and right arrow keys. Therefore this would require them to use left and right for both horizontal and diagonal, and up and down for vertical. All participants learned this response criteria quickly and no problems were observed during testing. Incorrect key responses were diasabled to help consolidate this response criteria.

Binning Data
Due to the large amount of SOAs, online data fitting of Session 1 was binned according to the total number of trials, and derived by the following formula dividing the range of SOAs by the square root (rounded up) of number of trials (n) in order to determine the number of bins rounded to the nearest whole number (t):

Reaction Times
For session 1, manual reaction times (RT) were recorded as the time difference between the onset of the first target and the time of the button press response. This onset was used to ensure consistency between reaction times despite varying SOAs. As the information required for making a choice is technically already available after the onset of one target. It is possible that reaction time data can be used to estimate certainty even in the absence of speeded responses in task requirements (Weiß & Scharlau, 2011). Therefore, a simple linear regression was performed to see if there was any influence of SOAs on reaction time. No significant correlation (r 2 = < 0.01, p = 0.49) was found