Effort adjustments come with a cost additional to the inherent cost of effort

Many theories on cognitive effort start from the assumption that cognitive effort can be expended at will, and flexibly up- or down-regulated depending on expected task demand and rewards. However, while effort regulation has been investigated across a wide range of incentive conditions, few investigated the cost of effort regulation itself.

Role: Lead Researcher

Methods: Computerized online experiment, online survey administration, A/B Testing

Tools: JsPsych (JavaScript), Prolific, R

Data Collection

I coded computerized online experiments in jsPsych (de Leeuw, 2015), and piloted and ran these experiments on Prolific.com (final experiment Ns = 45, 49, 48, 50). During the experiment, research volunteers completed a series of Random Dot Motion tasks (visual discrimination task: respond to the direction of coherently moving dots; Experiments 1 & 2) or Stroop tasks (cognitive conflict task: respond to the ink color of the word, ignoring the meaning of the word; Experiments 3 & 4), with varying cash incentive and cognitive load which was cued before the task. The cue of reward and load pertained to either one task trial (short prediction horizon: more frequent effort regulation), or six (long prediction horizon: less frequent effort regulation).

Data Analysis & Synthesis

To investigate the effect of the cognitive load, reward condition, and the frequency of effort level updating on effort, reaction time (ms from task onset to key press) and accuracy (correct: TRUE or FALSE) data were fitted with logistic and linear mixed-effects regression models in R to account for the nested nature of behavioral data within volunteers. 

To explore the effects of individual differences on the manipulation, questionnaire data were included in behavioral data models as predictor variables. The traits of interest were reward responsiveness (subscale of BIS-BAS; Carver & White, 1994) and affinity for cognitive challenges (Need For Cognition, short version; Cacioppo, 1984).

Results and Insights

Across visual discrimination and cognitive conflict tasks, reward prospect influenced effort adjustments only when the prospect pertained to multiple task attempts, suggesting the existence of a cost to effort regulation itself: if the task context keeps changing, keeping effort level stable might be the best strategy to optimize task performance.

 

Within the visual discrimination and cognitive conflict tasks, effort adjustments manifested only in accuracy or only in reaction times, respectively. The cognitive processes these tasks tap into might explain this discrepancy: the more data-limited random dot motion task is dependent on visual evidence accumulation, constraining reaction time adjustments. Conversely, the resource-dependent Stroop task relies on cognitive conflict resolution. Reaction time is thus more likely to reflect internally controlled processing speed.

 

Higher reward responsiveness was associated with larger reward-related performance adjustments, while higher affinity for cognitive challenges was linked to larger reward-related performance adjustments in the hard condition. The latter finding aligns with the idea that those who enjoy cognitive challenges experience effort to have a lower inherent cost.

 

  • One-off reward prospect might not lead to a boost in task engagement if future task reward and demands are volatile and unpredictable

  • People experience a cost to effort regulation, and lower effort adjustment frequency can compensate for this cost 

Click here to access the full research report