Electrophysiological and diffusion model parameter correlates of cognitive control in normal aging

• ¹ University of Newcastle, Australia

Increasing adult age is associated with changes in performance on some executive control tasks. Task-switching paradigms tend to show consistent age-related increase in mixing cost (RT on single task blocks vs. repeat trials within switch blocks) but less consistent age effects on switch cost (RT on repeat vs. switch trials in switch blocks). We examined the mechanisms underlying these effects in a large normative sample using cued trials switching paradigm. The excellent time resolution of event-related potentials (ERPs) was combined with diffusion model parameters of decision making in order to elucidate the cognitive processes that underlie age-related changes in task-switching performance. ERPs show consistent patterns of differences between switch and repeat waveforms time-locked to the onset of preparation (e.g., cue-locked) and stimulus-driven processes. Evidence accumulation models posit that, in 2-choice decision paradigms, the behavioral end-product (mean RT for correct responses and error rate) results from the combined effect of a number of latent variables. These contribute to the decision process (e.g., rate of evidence accumulation and response criterion) and to non-decision processes (e.g, stimulus encoding and response activation, and in the current context, cue encoding and task-set reconfiguration). We compared behavioral, diffusion and ERP measures of task-switching across a wide age range (18-80years, n=95). Participants were divided into 4 age groups matched on IQ, semantic fluency and digit span scores and were trained to switch between a letter classification and a digit classification task. Cue-stimulus interval (CSI: 150, 1000ms) was manipulated to assess age effects on task-set reconfiguration. Each task was mapped to two cues and all stimuli were incongruent. Mean RT showed the expected age-related increase in mixing cost only in 60-80 year olds. Diffusion parameters indicated a number of age-related changes in rate of evidence accumulation and response criterion related to RT mixing cost emerging in 46-59 year olds. These were accompanied by age-effects on both cue-locked and stimulus-locked ERPs that emerged even earlier (30-45 years), suggesting the early emergence of compensatory processes. Although mean RT switch cost was not affected by age, diffusion parameters and ERPs showed a number of age-related changes in the processing of switch trials. The older groups (45-59y & 60-80y) showed evidence of less efficient encoding and/or reconfiguration for switch than repeat trials. Younger adults prepared more efficiently for an impending switch trial, even at the very short preparation interval. These data indicate that diffusion model parameters and ERP measures are valuable and complementary approaches that illuminate the underlying processes that contribute to the age-related cognitive changes. The results suggest early emergence of strategic changes in task performance that efficiently compensate for less efficient information accumulation. Implications for models of task-switching and cognitive aging will be discussed.