The role of working memory in eye-gaze cueing Anna S. Law, Liverpool John Moores University Stephen R. H. Langton, University of Stirling Introduction Observers respond more quickly to stimuli that appear on the side of a computer screen that is being gazed at by a centrally presented face (valid trials), than to stimuli appearing at the opposite side (invalid trials). This eye-gaze cueing effect was initially thought to be highly reflexive (e.g., Friesen & Kingstone, 1998), but recent research has suggested that top-down attentional control may be involved (Koval & Everling, 2005; Vecera & Rizzo, 2006). Here we used a dual-task procedure, where participants attempted a standard gaze-cueing task alone or with a secondary task. If gaze-cueing depends on attentional control processes in working memory (WM), the effect should be modulated by WM load. Method Two groups of 20 undergraduate participants attempted a gaze-cueing task with a simultaneous digit rehearsal task (Lavie & de Fockert, 2006). One group had a high WM load (5 digits); the other had a low load (1 digit). Participants in each group did 2 blocks in counterbalanced order: one single task (gaze cueing only) and one dual task (gaze cueing + digit task). Results The RT data from the gaze-cueing task were analysed separately for each level of SOA, using a mixed ANOVA with 2 levels of gaze (valid vs. invalid), 2 levels of digit-load group (high load vs. low load) and 2 levels of demand (single task vs. dual task). There was a significant main effect of gaze validity at both 100ms SOA, F(1,38) = 6.43, p < 0.05, and 500ms SOA, F(1,38) = 10.54, p < There was no effect of gaze validity at 1000ms SOA, F<1. There was no effect of group at any SOA, and no interactions, all Fs <1. There were significant effects of dual task demand at 100ms SOA, F(1,38) = 56.37, p < 0.001, 500ms SOA, F(1,38) = 23.74, p < 0.001, and 1000ms SOA, F(1,38) = 33.70, p < However, there were no significant interactions between gaze validity and dual task demand (all Fs<1). Gaze cueing was not modulated by working memory load. Fig 3: Mean RT at 500ms SOA, collapsed across digit-load group Fig 1: “Valid” dual task trial for high load group For the low digit-load group, only 1 digit was displayed here The SOA between gaze cue and target was 100ms, 500ms, or 1000ms For the low-load group, a question mark was displayed Participants localised target with “left” and “right” button press responses Participants had to recall the number that came next in the sequence (high load) or in numerical order (low load) Fig 2: Mean RT at 100ms SOA, collapsed across digit-load group Fig 4: Mean RT at 1000ms SOA, collapsed across digit-load group Conclusions Working memory load caused an overall slowing of reaction times. However, the magnitude of the gaze-cueing effect remained the same under single and dual-task conditions. This could be seen as consistent with the traditional view that gaze cueing is highly reflexive, and not mediated by top-down control. A further experiment will use an executively-demanding secondary task, as this may be more disruptive to gaze-cueing. References Friesen, C.K., & Kingstone, A. (1998). The eyes have it! Reflexive orienting is triggered by non-predictive gaze. Psychonomic Bulletin & Review, 5, Koval, M. J., Thomas, B. S., & Everling, S. (2005). Task-Dependent Effects of Social Attention on Saccadic Reaction Times. Experimental Brain Research, 167, Lavie, N., & de Fockert, J. (2005). The role of working memory in attentional capture. Psychonomic Bulletin & Review, 12, Vecera, S. P., & Rizzo, M. (2006). Eye Gaze Does Not Produce Reflexive Shifts of Attention: Evidence From Frontal-Lobe Damage. Neuropsychologia, 44, Accuracy in the digit task: There were no significant effects in this analysis except a main effect of digit-load group, F(1,38) = 25.02, p < Average error rate was 2.41% for the low load group and 8.125% for the high load group. 3 Significant gaze cueing effect, but no interaction with demand No significant gaze-cueing effect