关键词: 医疗决策, 概率忽视, 人口老龄化, 情绪, 助推

Abstract:

Despite normative frameworks emphasizing integrative processing of probability and outcome information, real-world medical decision making often suffers from probability neglect, in which individuals place disproportionately more weight to potential adverse outcomes while neglecting their likelihood in treatment choices. To address this challenge, the present research systematically investigates the probability neglect phenomenon, underlying mechanisms, and corrective interventions for probability neglect in medical contexts through a multi-phase, multi-method program encompassing behavioral, psychophysiological, neurostimulation, and nudge-based studies.

First, Experiments 1 and 2 establish the prevalence of probability neglect in two pivotal populations, chronic illness patients and older adults, via within-subject comparisons of customized medical versus monetary decision tasks. Prior to decision tasks, participants complete monetary evaluation tasks for a spectrum of side effect severities, enabling calibration of personalized trade-offs between severity and probability. In the medical decision tasks, each side effect’s probability is determined by participants’ willingness to pay (WTP) to avoid that effect; in the parallel monetary decision tasks, side effects are replaced by their WTPs. Analyses of choice patterns reveal systematic underweighting of probability information in medical decisions relative to monetary contexts. Building on this foundation, Experiments 3 and 4 incorporate process tracing: Experiment 3 deploys eye-tracking to record fixation count and duration on probability versus outcome information, alongside real-time pupil dilation measures. Experiment 4 uses the Mouselab program to record open-box frequencies and compute search-measure (SM) indices, disentangling option- versus dimension-focused information-acquisition strategies. Together, these studies validate the ubiquity of probability neglect and uncover its associations with attentional biases and emotional responses.

Second, to isolate emotional drivers of probability neglect, Experiments 5-7 manipulate affective processing through context variation and targeted regulation strategies. Experiment 5 extends medical tasks into the domain of health supplements, comparing choices between medications and supplements to probe differential emotional impacts. Experiment 6 incorporates side-effect duration as an additional factor, testing whether prolonged adverse experiences amplify emotional arousal and probability neglect, particularly in supplement scenarios. Crucially, Experiment 7 adopts a mediation framework in which participants are randomly assigned to one of three emotion-regulation conditions—emotion-focus, emotion-suppression, or control—via standardized instructional sets. Behavioral outcomes and process measures assess whether attenuation of emotional reliance restores balanced probability weighting. Findings from these experiments elucidate causal links between emotion intensity and probability neglect, guiding development of emotion-targeted interventions.

Third, to obtain objective neurophysiological evidence, Experiments 8 and 9 adopt a multi-modal approach. Experiment 8 again employs the Mouselab program while continuously monitoring heart rate (HR) and skin conductance level (SCL) to index emotional arousal during decision tasks. We hypothesize higher physiological arousal corresponds to stronger probability neglect in medical versus monetary contexts. Experiment 9 employs transcranial direct current stimulation (tDCS) targeting emotion-processing cortical regions. Participants receive anodal, cathodal, or sham stimulation, after which choice behavior and attention measures reveal whether enhancing or inhibiting emotion-related cortical excitability causally alters probability weighting. These neurostimulation data furnish robust causal evidence for the emotional basis of probability neglect.

Fourth, informed by mechanistic insights, Experiments 10 and 11 design and evaluate two innovative nudge-based interventions aimed at correcting probability neglect. The first intervention leverages big data analytics of over 1.1 million social-media posts to extract empirically validated emotion-regulation behaviors, which are distilled into daily micro-nudges delivered via brief digital prompts (<10 min engagement). The second builds on self-perception theory, implementing a virtual AI feedback paradigm in which participants complete simulated conversational tasks with an AI agent that reinforces their belief that they are not influenced by emotions during decision making. Both interventions run over one week across experimental (AI nudge, big-data nudge, emotion-suppression instruction) and control groups. Outcome measures include pre- and post-intervention assessments of probability neglect in medical decisions, longitudinal follow-ups with older adults and chronic patients, and qualitative interview data to refine intervention content. Anticipated results demonstrate sustained reductions in probability neglect and enhanced decision quality, validating scalable, noncoercive strategies aligned with Thaler and Sunstein’s nudge philosophy.

By integrating personalized behavioral paradigms with process-tracing, psychophysiological monitoring, and neurostimulation techniques, this study advances both theoretical and practical frontiers in medical decision research. Empirical validation of causal affective pathways enriches the Dual System Model of Medical Decision Making (DSM-M), while novel digital and AI-based nudges offer pragmatic tools for healthcare delivery and telemedicine integration. Future work will explore embedding real-time emotion-monitoring modules into telehealth platforms to enable just-in-time decision support, thereby strengthening patient autonomy and optimizing health outcomes.

Key words: medical decision making, probability neglect, aging population, emotions, nudging