The neural underpinnings of the normative decision-making processes: A scoping review

Social norms guide everyday interactions, yet the neural processes through which they shapechoices remain inconclusive. We conducted a preregistered scoping review (OSF DOI:10.17605/OSF.IO/G379N) to map the neural substrates of normative decision-making (DM)and evaluate whether evidence accumulation (EA) offers a plausible neuro-computationalaccount. To position normative processes along a representational gradient, we compared theneural signatures across perceptual, value-based, social, and normative DM.Following JBI and PRISMA-ScR guidelines, we screened PubMed and Scopus (1990–2024)for task-based human studies using fMRI/fNIRS, EEG, or TMS/tDCS, excludingdevelopmental, clinical, pharmacological, and resting-state designs. From 209 eligible studies,we charted task domains, manipulations, neural readouts, spatial locations, and temporaldynamics.Across domains, core decision circuits were consistently engaged with domain-specificweighting. In normative DM (n=54 fMRI/fNIRS), primarily the dorsolateral PFC was mostfrequently involved (~44%), followed by medial/ventromedial PFC (~37%), anterior insula(~22%), and anterior cingulate cortex (~22%). EEG findings showed multi-timescale dynamicsdominated by MFN/FRN and P3, with modulation by fairness, partner identity, and framing;early N1/EPN effects appeared mainly in social tasks. Perceptual studies reliably exhibited EAsignatures (e.g., CPP), whereas value-based and social tasks showed sparse ramping dynamics,and no normative study linked choices to neurally constrained EA models.Normative DM appears to rely on coordinated control–valuation interactions (dlPFC–vmPFC–insula–ACC), yet its moment-to-moment influence remains underspecified. We suggestneurally informed EA modeling constrained by trial-level neural markers as a promisingframework. Future work should manipulate norm salience and identity cues, define “normativeevidence,” and combine multimodal neural measures with preregistered model-validationpipelines.