Prevention Focus Relates to Performance on a Loss-Framed Inhibitory Control Task

Abstract

Information framing can be critical to the impact of information and can affect individuals differently. One contributing factor is a person's regulatory focus, which describes their focus on achieving gains vs. avoiding losses. We hypothesized that alignment between individual regulatory focus and the framing of performance feedback as either gain or loss would enhance performance improvements from computer-based training. We measured participants' (N = 93) trait-level regulatory focus; they then trained in a go/no-go inhibitory control task with feedback framed as gains, losses, or control feedback conditions. Some changes in performance with training (correct rejection rate and response time) were consistent with regulatory fit, but only in the loss-framed condition. This suggests that regulatory fit is more complex than cursory categorization of trait regulatory focus and feedback framing might indicate. Regulatory fit, feedback framing, and task affordances should be considered when designing feedback or including game-like feedback elements to aid computer-based training.

Keywords: feedback; gamification; go/no-go; individual differences; inhibitory control; personalization; regulatory focus; training.

FIGURE 1

Timeline, point structure, and stimuli. (A) Single trial timeline of a go trial and the three kinds of feedback (gain framed, loss framed, and no points) for a correct response to a go stimulus. Each trial lasted 1.0 s and there was a 0.5 s inter-trial interval (ITI). (B) The curve relating response time relative to stimulus onset to the number of points achieved. (C) The stimuli for threat (go) and non-threat (no-go).

FIGURE 2

Prevention and promotion subscales for the regulatory focus questionnaire by experimental condition. Marginal plots show Gaussian smoothed histograms for prevention (top) and promotion (right) strengths to illustrate the relative frequencies of those strengths.

FIGURE 3

Time-courses of average training task performance. Participants were assigned to the condition that aligned (match, n = 37) or misaligned (mismatch, n = 24) with their predominant regulatory focus (Ctrl: no-points control condition, n = 31). Performance is summarized with (A) hit rate, (B) 20% trimmed mean response time for go stimuli, and (C) correct rejection rate. Summaries are computed over groups of three consecutive blocks. Shading shows a bootstrapped 95% confidence region.

FIGURE 4

Distributions of accuracy scores for the four kinds of stimuli in the transfer task. Error bars show a bootstrapped 95% CI for the mean computed on logit-transformed accuracy. Colors indicate the training condition, with match indicating those participants who were assigned to the condition that aligned with their predominant chronic regulatory focus (n = 37), mismatch indicating those assigned to the condition that misaligned with their predominant chronic regulatory focus (n = 24), and the control condition which involved no points (n = 31).

FIGURE 5

Change in correct rejection rate varies with prevention score. Change in log odds correct rejection (CR) is shown vs prevention score for (A) gain condition, n = 31, (B) loss condition, n = 30, and (C) control condition, n = 32. Solid lines show the expected mean, and dashed lines show a 95% confidence region around the expected mean. (D) Expected means for all three conditions.

FIGURE 6

Change in response time varies with prevention score. Response time (RT), summarized as the change in 20% trimmed mean, vs. prevention score for (A) gain condition, n = 31, (B) loss condition, n = 30, and (C) control condition, n = 32. Solid lines show the expected mean, and dashed lines show a 95% confidence region around the expected mean. (D) Expected means for all three conditions.

FIGURE 7

Initial correct rejection rate varies with prevention score. Initial log odds correct rejection (CR) is shown vs. prevention score for (A) gain condition, n = 31, (B) loss condition, n = 30, and (C) control condition, n = 32. Solid lines show the expected mean, and dashed lines show a 95% confidence region around the expected mean. (D) Expected means for all three conditions.