1. Participants seem to utilize predictive information to actively maintain both task sets when doing so aids performance. Decreased mixing and switch costs for predictable, equiprobable stimuli, the decrease of cue-related P3 amplitude from rare to equiprobable switches, and the absence of any P3 to informative cues for equiprobable, predictable switches all suggest that cue information is increasingly irrelevant to performing the task as switches become more probable and predictable. Although response conflict was highest after uninformative cues, smaller target P3s for predictable vs. unpredictable equiprobable switches were observed, suggesting that predictability allowed processing resources to be conserved. Participants may also have created a more global behavioral set encompassing both tasks [3]. Switch costs in the equiprobable condition might be best described as relatively “pure” costs of switching between active task sets rather than as proactive interference. Frequent change can be good for you: ERP evidence that task switch probability affects cognitive control Michael Bersick (1), Doreen Nessler (1), Ray Johnson, Jr. (2), and David Friedman (1) 1 Cognitive Electrophysiology Laboratory, NY State Psychiatric Institute, NY, NY;2 Department of Psychology, Queens College/CUNY, Flushing, NY Uninformative cue, predictable Informative cue, unpredictable Uninformative cue, unpredictable Informative cue, predictable 8 Discussion Introduction The cost of switching between two tasks is reduced by providing a cue warning of an impending switch. Cognitive control accounts of this effect claim that an informative cue allows for task-set reconfiguration prior to the actual switch [1]. Nevertheless, the fact that switch costs remain even after a long preparatory interval suggests that different task sets can interfere with each other [2]. However, such proactive interference might only be a problem when successive task sets must be reactivated. If switching is frequent, then keeping both task-sets active is a good strategy, assuming that little cost accrues from doing so. We examined how switch probability, switch predictability, and informative-cue availability interact to change the ease with which participants can maintain and access different task sets. Results Fig. 3: Informative cues reduced switch costs. In addition, when switches were equiprobable switch costs after uninformative cues were significantly reduced. Mixing CostsSwitch Costs Fig. 4: (b) Informative cues preceding equiprobable, unpredictable switches elicited a parietal, P3-like effect relative to pre-switch trials. (a) This effect was absent for informative cues on equiprobable, predictable switch trials. Fig. 5: (a) and (b) Informative cues preceding both predictable and unpredictable rare switches elicited much larger amplitude P3s than those present in the equiprobable condition (cf. Fig 4). Predictable, informative cue Unpredictable, informative cue Predictable, uninformative cue Unpredictable, uninformative cue Fig 2: Informative cues reduced mixing costs. Mixing costs decreased dramatically after uninformative cues in the equiprobable condition, but only when task order was predictable. 05001000 ms-500-1000 Cue-related P3 (a) Predictable Pz Single task Pre-switch Switch Pz Informative cue Uninformative cue Equiprobable Cz 6a 5001000 ms0 Equiprobable Cz 6b 4a Rare switch Cz 5001000 ms0 7b 7a Participants: Fifteen young adults (19 to 30 years; M = 24.6; SD = 3.8). Design: Digit (not 5) required response: more/less than 5? or odd/even? pure blocks: one task; mixed blocks: two tasks. Factors: (1) Ratio of switch (S) to stay or repeat (R) trials equiprobable; switch after 0, 1, or 2 trials: a-bb-aaa rare switch 1:3; switch after 2, 3, or 4 trials: aaa-bbbb-aaaaa (2) Cue-status: informative, uninformative ( Fig. 1b ) ; (3) Predictability Status: predictable, unpredictable Mixing costs = pre-switch RTs in mixed blocks - RTs in pure blocks Switch costs = switch RTs – pre-switch RTs mixed blocks EEG recording: 62 sintered Ag/AgCl electrodes; referred to averaged mastoids; continuous DC- 100Hz; 500 Hz sampling rate; Fig. 1a: ERP epochs. Methods Informative Uninformative 4 4 ? 1300800300 500 Target onset Cue onset 1000 ms Baseline for cue- related ERPs 0 Baseline for target-related ERPs 1a 1b Fz 200400 ms MFN Fig. 8: After uninformative cues, response conflict, as indicated by the amplitude of the MFN, was larger for unpredictable than predict- able, equiprobable switches. Fig. 6: After uninformative cues, the decrease in the amplitude of the target P3 was larger for unpredictable (b) than for predictable (a) pre-switch and switch trials. Fig. 7: (a) and (b): Target P3s to rare switch and pre- switch stimuli. (b) Unpredictable 4b 05001000 ms-500 -1000 Target P3 Pz Rare switch 5b 5a more/less than 5? odd/even? References [1] Monsell & Mizon (2006). JEP: HPP (32), 493-516. [2] Wylie & Allport (2000). Psychological Research (63), 212-233. [3] Schneider & Logan (2006). JEP: General (135), 623-640. RT Cz cue onsettarget onset