Introduction

The discovery that decision-making under uncertainty is susceptible to irrelevant contextual factors—such as whether outcomes are framed as gains or losses—has fundamentally challenged rational choice theory. Tversky and Kahneman's prospect theory elegantly explains these violations by positing an editing phase of deliberation wherein framing effects arise from asymmetric risk attitudes in gain versus loss domains. However, the cognitive mechanisms underlying frame sensitivity remain contested. Recent dual-process models propose that resource-demanding deliberative processes can be bypassed when attentional capacity is constrained, potentially increasing reliance on simpler heuristic judgments.

If frame effects reflect deliberative processing, imposing concurrent cognitive load should diminish or eliminate these effects. Conversely, if framing effects stem from automatic evaluative processes, load might enhance them. We tested these competing hypotheses by examining whether a competing memory task modulates the classic pattern of loss aversion in framing scenarios, with particular attention to the role of probability information in mediating this effect.

Method

Participants

Sixty-five undergraduate students (mean age 19.8 years, SD 1.5; 38 female) from a Canadian university participated for course credit. All participants were native English speakers with no known cognitive or attentional deficits.

Procedure

Participants completed a computerized decision task programmed in E-Prime, presenting 18 choice problems adapted from Tversky and Kahneman's classic 1981 Asian Disease Paradigm. Nine scenarios were presented in gain frames ("save X lives") and nine in loss frames ("X lives will die"). Within each frame, three probability levels were manipulated: high probability (p = 0.80), medium probability (p = 0.50), and low probability (p = 0.20). Each problem offered a choice between a certain option and a gamble. In the high-load condition, participants simultaneously maintained a 6-digit number in working memory, which they recalled at the end of the task. The low-load control condition presented the same decision problems without the memory load. Problems were presented in random order across two blocks (one per load condition, order counterbalanced).

Primary outcome measures were the frequency of risk-averse choices (selecting certainty) in gain frames and the frequency of risk-seeking choices (selecting the gamble) in loss frames. These responses define the framing effect. Secondary measures included response latencies and memory recall accuracy in the load condition (96% on average).

Results

A mixed-model ANOVA with frame (gain, loss), probability (high, medium, low), and load (full, reduced) as factors revealed a significant three-way interaction (F(2,63) = 4.18, p = 0.020). Frame effects were robust at high and medium probabilities in both load conditions (gain frame risk aversion: 72% and 68%; loss frame risk-seeking: 64% and 59%). However, at low probabilities, cognitive load reversed the framing effect: in the full-attention condition, participants remained risk-averse in gains (59%) and risk-seeking in losses (61%), consistent with classic patterns. Under load, this difference diminished (gain frame: 53%; loss frame: 52%; interaction F(1,63) = 5.73, p = 0.020).

Response latencies were significantly longer in the low-probability condition (M = 2,847 ms, SD = 1,106) than high probability (M = 1,924 ms, SD = 687; t(64) = 6.34, p < 0.001), suggesting that probabilistic judgment at lower thresholds demands greater deliberation. Load increased latencies overall by approximately 340 ms (F(1,63) = 22.65, p < 0.001).

Discussion

These findings support a selective process-switching account wherein deliberative reasoning is engaged proportionally to decision complexity. When probabilities are high and outcomes are concrete, both attentional conditions yield consistent frame effects, suggesting automatic valuation suffices. However, when probabilities are low and abstractly presented, individuals under full attention engage effortful recomputation of prospect value, maintaining frame-consistent choices. Cognitive load impairs this deliberative override, unmasking a more "objective" probability weighting that treats gains and losses more equivalently.

The results have practical implications for naturalistic high-stakes decisions—medical choices, financial planning, public health communications—wherein attentional constraints are common. Communicators attempting to promote health or safety behaviors may be counterintuitively more effective when framing messages for gains (emphasizing lives saved) in high-complexity scenarios, where load-induced processing reductions are most likely to occur. Future research should examine whether individual differences in working memory capacity moderate these effects and whether brief attentional recovery intervals can restore frame sensitivity.

References

  • Epstein, S., & Pacini, R. (1999). Some basic issues regarding dual-process theories. In S. Chaiken & Y. Trope (Eds.), Dual-process theories in social psychology (pp. 462–482). Guilford Press.
  • Kahneman, D., & Frederick, S. (2002). Representativeness revisited: Attribute substitution in intuitive judgment. In T. Gilovich, D. Griffin, & D. Kahneman (Eds.), Heuristics and biases: The psychology of intuitive judgment (pp. 49–81). Cambridge University Press.
  • Stanovich, K. E., & West, R. F. (2000). Individual differences in reasoning: Implications for the rationality debate. Behavioral and Brain Sciences, 23(5), 645–726.
  • Tversky, A., & Kahneman, D. (1981). The framing of decisions and the psychology of choice. Science, 211(4481), 453–458.
  • Tversky, A., & Kahneman, D. (1992). Advances in prospect theory: Cumulative representation of uncertainty. Journal of Risk and Uncertainty, 5(4), 297–323.
  • Weber, E. U., & Johnson, E. J. (2009). Psychology and behavioral decision making. In R. M. Dawes & R. Hastie (Eds.), Rational choice in an uncertain world: The psychology of judgment and decision making (2nd ed., pp. 297–322). Sage.