Poster #3 - Shifting Perceptions on Cognitive Flexibility in Childhood: A Diffusion Model Analysis

• In Event: Poster Session 13

Abstract

Flexibly and rapidly switching between task demands is a key cognitive skill. On cognitive tasks, response times (RTs) on trials requiring a rule switch are consistently slower than nonswitch trials, producing a “switch cost” between switch and nonswitch trials (Monsell, 2003). Unlike traditional methods, the diffusion model (DM) simultaneously accounts for cost effects in speed and accuracy, integrating them into a single set of performance metrics (Ratcliff & McKoon, 2008). DM analyses attribute slower and error-prone performance on switching tasks in children to reduced drift rate (less efficient accumulation of evidence), boundary width (less efficient speed/accuracy tradeoff), and Ter (more time for processes unrelated to decision-making) relative to adults (Weeda et al., 2014).
Whereas extant work has predominantly used dichotomous stimuli, the current study seeks to examine mechanisms driving switching between dimensional stimuli, which may more closely resemble information processing in real-life decisions (Kvam et al., 2022). We expect reduced speed and accuracy on switch trials and harder task-set trials, due to slower drift and longer Ter. Finally, we expect slower RT and worse accuracy on difficult trials.
Twenty-eight children aged 8-12 completed two computerized switching paradigms. In “Box,” children indicated whether 31-70 asterisks arranged in a 10x10 array appeared on the right/left of the fixation point or whether the array contained “many” or “few” asterisks. In “Arrows,” children indicated whether a central arrow pointed to the right/left or was drawn in a light/dark color (20%-80% dark grey). Each trial was preceded by a 350 ms cue-stimulus interval indicating the decision to be made. All stimuli were dimensional to allow for variation in difficulty. To obtain DM parameters, RT and accuracy on 192 trials across 3 blocks were entered into the Fast-DM modeling program and permitted to vary by task, switch, and trial difficulty (Voss & Voss, 2007).
As expected, Switch trials and specific Tasks (number/color) produced slower RTs and lower accuracy (Switch: all p<0.006; Task: all p<0.008) due to slower drift rate and longer Ter (all p<0.05; see figures). Specifically, switching from Number/Color Tasks to Side Task yielded slower RTs (SwitchxTask interaction, all p<0.004) . In “Box,” this interaction was due to slower drift (F(1,27)=31.05,p<0.001). There were no other interactions.
Difficult decisions slowed RT (all p<0.002) and lowered accuracy (all p<0.009), particularly when switching from a Number/Color Task to the Side Task (TaskxDifficulty interaction, all p<0.006). In “Box,” this interaction produced worse accuracy for the difficult trials (F(1,27)=38.79,p<0.001). In “Arrows,” there was a SwitchxTaskxDifficulty interaction for RT (F(1,27)=22.52,p<0.001).
Children were slower and less accurate on Switch trials due to slower drift rate and Ter. Dimensional stimuli produced similar findings to previous work with dichotomous stimuli, and suggest that the mechanism responsible for switch are generalizable. Asymmetric switch costs were also observed in which switch costs were larger when shifting from the harder to easier trials. This phenomenon has previously been observed, but the cause is unclear (Schneider & Anderson, 2009). Future work is encouraged to further examine this phenomenon in tasks with sufficient trial numbers to allow DM analyses.

Authors

• Christina Olena Hlutkowsky, The Pennsylvania State University
  Presenting Author

• Jason Scot Feldman

• Elsie Xu Yan, Pennsylvania State University, University Park

• Cynthia Huang-Pollock, Pennsylvania State University