Four converging lineages — the diffusion model & random-dot motion, face/emotion-DDM, gaze-direction DDM, and continuous flash suppression — from 1978 to this design contract. Each milestone links to its DOI where known.
Psychological Review. Introduced the continuous random-walk (diffusion) model of decision
making: evidence accumulated as a Wiener process with drift, terminating at one of two absorbing
boundaries. This is the mathematical framework (v, a, t₀) later used to analyze
both RDM and face 2AFC data.
Journal of Neuroscience. Trained monkeys on random-dot direction discrimination; each animal individually calibrated to threshold coherence — establishing the per-subject coherence-calibration paradigm this design's RDM arm inherits.
doi.org/10.1523/jneurosci.08-06-02201.1988 · Bibliography entryNature. Simultaneously measured psychophysical coherence thresholds and single MT-neuron sensitivity, showing individual neurons can rival behavioral thresholds. Origin of the %-motion-coherence construct used throughout the RDM literature.
doi.org/10.1038/341052a0 · Bibliography entryJournal of Neuroscience. Compared psychophysical and single-neuron MT sensitivity for direction discrimination in the same animals on the same trials; individual MT neurons matched psychophysical threshold.
doi.org/10.1523/jneurosci.12-12-04745.1992 · Bibliography entryJournal of Neuroscience. Numerical model of how pooled MT signals feed downstream (LIP) decision circuitry, generating predictions about psychometric functions — the neural-to-behavioral bridge that the RDM-arm DDM analysis formalizes.
doi.org/10.1523/jneurosci.16-04-01486.1996 · Bibliography entryVision Research. Jointly estimates psychometric threshold and slope by selecting each stimulus to minimize expected posterior entropy. Along with QUEST (Watson & Pelli 1983), this is the adaptive-calibration tradition this design imports wholesale for face-task titration — a manipulation with 0/19 precedent in the face-DDM literature.
doi.org/10.1016/s0042-6989(98)00285-5 · Bibliography entryJournal of Neurophysiology. LIP neural activity ramps toward the ultimate choice, reflecting accumulation of sensory evidence at multiple coherence levels — the physiological substrate of the drift-diffusion account of RDM decisions.
doi.org/10.1152/jn.2001.86.4.1916 · Bibliography entry · PDFJournal of Vision. Unified diffusion-model theory linking coherence to both RT and
accuracy via μ′ = k·x, with sensitivity modeled as k ~ N(21, 6) between subjects (range ≈ 9–28).
This is the direct, well-established anchor for the RDM arm of this design.
Nature Neuroscience. Introduces CFS: distinct images flashed at ~10 Hz to one eye suppress a stimulus in the other eye far longer than standard binocular rivalry. The 10 Hz Mondrian-mask dynamics this design's graded binocular masking borrows from — though Part 1 is deliberately pre-CFS (binocular, not dichoptic).
doi.org/10.1038/nn1500 · Bibliography entryPsychonomic Bulletin & Review. Three closed-form equations turning mean RT, RT
variance, and accuracy into v, a, t₀. The fast per-subject screening estimator
this design uses — despite having 0/19 precedent among
the face-DDM corpus.
Emotion. Diffusion-model re-analysis of a threat-word lexical-decision task: high-anxious participants show a drift-rate advantage for threat words that mean RT/accuracy missed entirely. Foundational demonstration that DDM sees what raw RT can't — the methodological template this design's face arm follows, even though the stimuli are words, not faces.
doi.org/10.1037/a0019474 · Bibliography entry · PDFCognition & Emotion. 67 models × 8 expressions × 3 gaze directions × 5 camera angles, validated and free for academic use. The stimulus set this design uses — justified on its own validation merits and its 3 gaze directions (needed for V1/V3). RaFD is also used by our anchor paper (Williams 2023, Study 1), so it is not novel to the corpus.
doi.org/10.1080/02699930903485076 · Bibliography entry · PDFFrontiers in Human Neuroscience. Critical evaluation of the b-CFS paradigm using upright vs. inverted face detection; CFS amplifies rather than creates the upright-face advantage. Central methodological caution for any future b-CFS extension of this project.
doi.org/10.3389/fnhum.2011.00167 · Bibliography entry · PDFFrontiers in Neuroinformatics. Python toolbox for hierarchical Bayesian DDM estimation; estimates individual and group parameters simultaneously and needs fewer trials/participant than classical methods. The plurality choice among face-DDM papers that name a method (4/6) and the confirmatory half of this design's EZ+HDDM pairing.
doi.org/10.3389/fninf.2013.00014 · Bibliography entry · PDFRoyal Society Open Science. Direct left/right gaze-direction 2AFC with head and eye angle independently, parametrically varied — the best task-structure analog for this design's V1 gaze arm — but reports thresholds in degrees only; no DDM/v/a/t₀ fit at all.
doi.org/10.1098/rsos.180885 · Bibliography entryCognition. Diffusion-model analysis of a face-in-the-crowd search task: drift rate larger and non-decision time shorter for angry/happy faces than energy-matched anti-expressions. Numeric magnitudes remain paywalled — see Gaps §5.
doi.org/10.1016/j.cognition.2022.105235 · Bibliography entryScientific Reports. Two pre-registered studies (N=228, N=264): masking the lower (expressive) face region lowers drift rate v (b=−0.38→−0.65 collapsed, up to −1.12 for individual emotions). The strongest single anchor in the corpus for this design's central hypothesis, though masking there is binary, not graded.
doi.org/10.1038/s41598-023-35381-4 · Bibliography entry · PDFQuarterly Journal of Experimental Psychology. First DDM decomposition of gaze-cueing behavior across three datasets (N=171, 139,001 trials); finds gaze shifts non-decision time far more than drift (v inclusion 3–13% across datasets) — a caveat for using it as a drift-rate anchor for this design's gaze arm.
doi.org/10.1177/17470218231181238 · Bibliography entryA cluster of studies extends emotion-DDM to new populations and manipulations: Ozturk et al. 2024 (threat-uncertainty cueing, HDDM), Nan et al. 2024PDF (testosterone & anger sensitivity, pharmacological RCT), Haller et al. 2024 (developmental happy–angry morph continuum), Klein & Todd 2024PDF (race & weapon identification), Schreiber et al. 2025PDF (borderline personality symptoms), Nagrodzki et al. 2025 (incidental attention to anger), Evans et al. 2025 (approach–avoidance, pre-registered replication). None closes the graded-masking + strict-2AFC + full-DDM gap on either axis — see the zero-precedent flags.
Combines the RDM/DDM anchor (Palmer, Huk & Shadlen 2005PDF), the masking→drift direction (Williams 2023PDF), RaFD (Langner 2010PDF), and the psychophysics calibration tradition (Kontsevich & Tyler 1999) into a single within-participant design — a combination with zero precedent in the reviewed literature on either the gaze or emotion axis. See The Design and Golden-Standard Comparison.