The study, conducted by Can Cui of Xuzhou Medical University and Northeast Normal University, along with Yiting Long, Chu Zhang, Yuan Yuan, and Yingjie Jiang, tackled a question that has lingered at the edge of creativity science: do reward effects generalize across gain and loss framing? Most prior work has explored how incentives shape creative cognition when people are told they can earn something. Far less attention has been paid to what happens when the same incentive is presented as a threat, something that could slip away if performance falls short. Given that real-world creative work, from scientific grants to artistic commissions, is often framed in terms of what might be lost rather than gained, the gap is more than academic.

To probe this, the team ran two experiments that manipulated two key properties of rewards: whether the money was real or merely hypothetical, and whether the amount was high or low. Participants solved problems under loss framing, meaning the reward was described in terms of potential loss rather than gain. The researchers then compared these new loss-frame results against previously published gain-frame datasets collected with the same tasks and the same reward manipulations, allowing a rare direct look at how framing itself reshapes motivational effects on cognition.

Two distinct creative tasks anchored the investigation. The first was the compound remote associates task, or CRA, a classic measure in which participants must find a single word that links three seemingly unrelated cue words. Solving these problems often involves a flash of insight, a sudden restructuring of how the elements relate. The second was a Chinese character chunk-decomposition task, or CD, in which participants must break apart unfamiliar or ill-structured character configurations into their component strokes or radicals. Crucially, the CD task distinguishes between ordinary decompositions, which rely on familiar radical-level units, and novel decompositions, which require breaking free of learned representations down to the level of individual strokes. This distinction gave the researchers a window into two flavors of creative processing: retrieving known structures versus constructing genuinely new ones.

The results revealed a striking asymmetry between the two tasks. For the CRA task, no interaction involving framing was detected for solution rates, suggesting that whether rewards were framed as gains or losses did not alter how many problems participants solved. However, a reward type by frame interaction did emerge for response times, hinting that framing subtly shifted how quickly people worked, even when their success rates held steady. Insight-driven word problems, it seems, are relatively robust to how the stakes are described.

The chunk-decomposition task told a very different story. Here, the researchers uncovered a significant three-way interaction among reward type, reward magnitude, and framing, driven almost entirely by performance on novel, stroke-level decompositions, the very problems that demand departing from familiar representations. For these novel challenges, the reward pattern differed pronouncedly between the loss-frame sample and the previously collected gain-frame data. In contrast, for ordinary, radical-level decompositions that could be solved with well-practiced perceptual chunks, the framing-related estimate was close to zero. In other words, loss framing selectively rewrote the reward rules only for the most representationally demanding form of creative thinking.

Within the loss-frame sample itself, the pattern of results was counterintuitive. Real rewards of high magnitude produced better performance than either hypothetical high rewards or real low rewards, consistent with the intuition that tangible, substantial incentives motivate effort. Yet the highest overall performance came from the combination of hypothetical rewards and low magnitude, a pairing that on the surface should have been the least motivating. This paradox suggests that under loss framing, the psychological meaning of a reward is not simply a function of its size or reality, but of how these properties interact with the threat context to shape cognitive strategy.

To make sense of the findings, the authors turned to a metacontrol framework, a theoretical account in which cognition continually balances two competing modes: persistence, which stabilizes current representations and drives focused effort, and flexibility, which loosens constraints and permits restructuring. In this view, reward properties act as biases on the persistence-flexibility balance. Real, high-magnitude rewards may push the system toward persistence, which helps with problems requiring sustained application but can hinder the representational upheaval needed for novel decompositions. Hypothetical, low-magnitude rewards, by contrast, may leave the system freer to explore, an advantage precisely when problems demand breaking apart familiar structures. Loss contexts, the researchers propose, may additionally alter how novel representations are processed in the first place, adding a further layer of modulation beyond simple motivational intensity.

The authors are careful about the limits of their design. Because the gain-frame and loss-frame data came from independent samples rather than participants tested under both framings, the observed differences cannot be attributed uniquely to framing itself. Sample-level differences in ability, motivation, or strategy could contribute to the cross-sample patterns. Still, the specificity of the effect, confined to novel decompositions and absent from ordinary ones and from CRA solution rates, argues against a generic explanation and lends weight to the idea that framing genuinely interacts with the representational demands of the task.

The implications reach beyond the laboratory. Educational settings, workplace innovation programs, and incentive structures of every kind routinely deploy rewards without considering whether they are framed as gains to be won or losses to be avoided. If the findings hold up, the same bonus could enhance routine problem solving while quietly undermining the kind of thinking that produces genuinely original solutions, or vice versa depending on the framing. Motivational framing, the researchers conclude, may be a boundary condition on reward effects in creative problem solving, a factor that determines when incentives help and when they hurt. For anyone designing systems meant to cultivate human creativity, from classrooms to research labs, the message is that how you say it may matter as much as what you offer.

Subject of Research: How loss versus gain framing of rewards modulates creative problem solving performance

Article Title: Loss framing selectively reorganizes reward effects on creative problem solving

Article References: Cui, C., Long, Y., Zhang, C., Yuan, Y., & Jiang, Y. (2026). Loss framing selectively reorganizes reward effects on creative problem solving. BMC Psychology. https://doi.org/10.1186/s40359-026-05801-x

Image Credits: AI Generated

DOI: 10.1186/s40359-026-05801-x

Keywords: creative problem solving, reward, loss framing, gain frame, chunk decomposition, compound remote associates, metacontrol, creativity, cognitive psychology, motivation, insight, BMC Psychology