What Happens during the Stimulus Onset Asynchrony in the Dot-Probe Task? Exploring the Role of Eye Movements in the Assessment of Attentional Biases

Bermeitinger C (2013) What Happens during the Stimulus Onset Asynchrony in the Dot-Probe Task? Exploring the Role of Eye Movements in the Assessment of Attentional Biases. PLoS ONE 8(10): e76335. doi:10.1371/journal.pone.0076335 What Happens during the Stimulus Onset Asynchrony in the Dot-Probe Task? Exploring the Role of Eye Movements in the Assessment of Attentional Biases Kalina Petrova 0 Dirk Wentura 0 Christina Bermeitinger 0 Philip Allen, University of Akron, United States of America 0 1 Department of Psychology, Saarland University , Saarbru cken, Germany , 2 Department of Psychology, University of Hildesheim , Hildesheim , Germany The dot-probe paradigm is one of the most often used paradigms to investigate attentional biases towards emotional information. However, a large number of the dot-probe studies so far used a long stimulus onset asynchrony allowing for eye movements to occur, which might increase the error variance. This study aimed at addressing this methodological issue by varying the instructions with regard to the gaze behavior and calculating the reaction time (RT) bias score (i.e., RTs for targets presented at the location of the emotional compared to the neutral stimulus) separately for trials with eye movements and trials without eye movements. Results of Experiment 1 (using typical instructions, i.e., instructions that are lenient with regard to eye movements) showed an RT bias, but only in the trials without eye movements The overall RT bias (calculated ''blind'' for eye movements) was non-significant. In Experiment 2, stricter instructions and small changes in the procedure led to a sharp decrease in the number of eye movements, such that both the RT bias in the trials without eye movements as well as the RT bias across all trials was significant. - A frequently used paradigm to investigate selective attention towards emotional information, particularly in experimental psychopathology, is the dot-probe paradigm (see [1], for a review). In this paradigm, participants are typically presented with two stimuli, one neutral and one emotional, simultaneously side by side for a brief period of time. Subsequently, a probe appears either in place of the neutral stimulus (invalidly cued condition) or in place of the emotional stimulus (validly cued condition), and participants have to categorize it according to a spatially unrelated dimension (e.g. [2]). The rationale in this paradigm is similar to the one in the exogenous spatial cueing paradigm, namely that participants should be faster in trials in which the probe appears in the validly cued (viz. attended) location than in trials in which the probe appears in the invalidly cued (viz. unattended) location (see [3]). Thus, faster reaction times (RTs) in validly cued than in invalidly cued dot-probe trials (i.e., an RT bias) are interpreted as emotionrelated attentional bias. However, there is one important caveat. Compared with cueing studies in basic cognitive research, a rather long stimulus onset asynchrony (SOA) of cues and probes (i.e., 4001250 ms) has typically been used in dot-probe studies. Therefore, the RT bias could be attributed to either enhanced vigilance or delayed disengagement of attention (e.g. [4,5]). Furthermore, basic research with the cueing paradigm indicates that such long SOAs are associated with longer RTs for validly cued trials compared with invalidly cued or neutral trials (inhibition-of-return effect; [6]). Moreover, such long SOAs are sufficient for the programming and execution of at least one eye movement. Thus, the dotprobe paradigm can measure covert shifts of attention (i.e., changes in the focus of attention that took place without eye movements) as well as overt orienting of attention (i.e., eye movements). Whereas, however, in basic cueing studies either very brief SOAs (precluding eye movements) are used, or only trials in which participants made no eye movements enter into the analyses in order to make sure that the effects found are not due to overt orienting, in dot-probe studies no differentiation is typically made between covert and overt attention. From the perspective of someone interested in attentional biases towards emotional stimuli one might ask why it is important to control for the occurrence of eye movements in the dot-probe task given that gaze shifts closely follow shifts in covert orienting (e.g. [7]). First, previous studies have shown prolonged reaction times when the responses were given shortly after a saccade compared to the condition in which no saccade preceded the response (e.g. [8]; see also [9]). This finding has been attributed to an inhibitory interaction between the saccades and the manual responses in the afferent system or at the motor control site [8]. Thus, one might argue that in dot-probe trials in which an eye movement occurred just before the probe onset, the reaction time effects diminish and the dot-probe effects are solely based on the trials in which no eye movements occurred. Alternatively, given the strong relationship between covert and overt attention, one might argue that in trials in which an eye movement was directed towards the emotional stimulus a positive RT bias score results (i.e., faster reaction times in validly cued trials compared to invalidly cued trials), whereas in trials in which an eye movement was directed towards the neutral stimulus a negative RT bias score results (i.e., faster reaction times in invalidly cued trials compared to validly cued trials). In any given case, it is important to investigate which trials underlie the RT bias effect (i.e., those in which no eye movements occurred or those in which an eye movement was directed towards the emotional stimulus) or whether both types of trials lead to similar effects. Second, long SOAs (i.e., .300 ms) are usually associated with the phenomenon of inhibition of return (see above). Undoubtedly, there is strong evidence suggesting that covert attention is biased towards emotional information with SOAs below 100 ms, as the rare studies using such short SOAs indicate (e.g. [2,10]). There is also some evidence suggesting that covert orienting is biased towards emotional information even with long SOAs. For example, [9] found that some participants made a lot of eye movements whereas others only rarely directed their gaze towards the cues. Those who made few eye movements had faster RTs than those who made a lot of eye movements. Whether or not participants made a lot of eye movements did not modulate the effect of trait anxiety on the the RT bias score (i.e., larger RT bias score in the high and medium anxious than in the low anxious participants). Thus, the authors concluded that the RT bias was independent of the extent of overt orienting. However, no study has yet separated the effects of overt and covert orienting on the RT bias in the dot-probe task at the level of individual trials. Third, currently there is evidence that eye move (...truncated)