5. Borderline case policy

These areas required explicit policy decisions. Section 5.1 was resolved; the remaining five are open. The variation audit applied the default rules listed below:

5.1 Emotional variants

Decision: The EMOT DROP category was retired. All attested Emotional-X paradigms are kept as variations under their parent tasks. No emotional variations are promoted to top-level tasks (decided).

What this means in practice: The following emotional variations are kept as variations (not top-level tasks): Emotional Anti-Saccade, ANT with Emotional Stimuli, CPT with Emotional Distractors, Emotional Directed Forgetting, Emotional Flanker, Emotional Navon, Emotional Oddball, Emotional Simon Task, Emotional Go/No-Go, Emotional N-Back, Emotional TNT, Emotion-Induced Blindness, Emotional Attentional Blink. The Emotional Stroop Task, Emotion Regulation Task, and Facial Emotion Recognition Task remain top-level tasks (they were always independent entries, not promoted from variations).

Rationale: Emotional content does change what the participant experiences, and these are attested paradigms in the literature. Keeping them as variations rather than promoting them avoids proliferating task entries for every stimulus-domain cross.

5.2 Proportion-congruent manipulations

Current rule: KEPT for Flanker, Stroop, and Simon tasks because proportion-congruent manipulations change the cognitive control context substantially (participants adjust their strategies based on conflict frequency).

Open question: Could be argued as a design choice (DESG) since it’s an experimenter-controlled parameter.

5.3 Child/developmental versions

Current rule: KEPT when the procedure is genuinely simplified or uses different stimuli that change the task. DROPPED when only stimuli are swapped for age-appropriateness without changing the procedure.

Currently kept child/developmental variations (5): Child ANT (fish stimuli, simplified flanker), Developmental / Child IAT (picture-based categories), Iowa Gambling Task Child Versions (simplified decks), RMET Child Version (reduced item set, simplified vocabulary), Child-Adapted SART (modified timing and stimuli).

Open question: Should there be a uniform rule for all child versions?

5.4 Dual-task add-ons

Rule: Uniformly DROPPED. Adding a concurrent secondary task (working memory load, articulatory suppression, etc.) is a design choice (DUAL), not a task variation, regardless of how widely used the combination is. No exceptions for “well-established” dual-task pairings - popularity does not change the structural argument.

Applied: All DUAL-coded entries were removed in the original audit. The one borderline survivor from that pass - Concurrent Cognitive Load (Affective Picture Viewing Task) - was dropped for consistency.

5.5 Confidence ratings / subjective reports

Rule: Adding a confidence rating or subjective report to an otherwise unchanged task is a measurement add-on (MEAS), not a variation. The fact that it adds a participant response to the event stream is not sufficient - by that logic any bolted-on scale would create a variation. Confidence-rating add-ons are uniformly DROPPED.

Exception: A confidence-related variation is kept only when the confidence judgment defines a recognized named paradigm whose construct of interest is the metacognitive correspondence, not the base task’s performance. The sole current example is the Confidence-Accuracy Paradigm (Heartbeat Detection Task), where Garfinkel et al. (2015) defined a three-dimensional model of interoception in which confidence-accuracy correspondence is one of the core measured dimensions.

Dropped in the original audit (6): AGL with Confidence Ratings, ROC Analysis (Old/New Recognition), Confidence Judgment Version (Random Dot Kinematogram), ROC Analysis (Remember/Know), Backward Masking with Confidence Ratings (Visual Masking), Source Confidence Ratings (Source Memory). Dropped (2): Confidence-Rated Recognition (Old/New Recognition), Remember/Know with Confidence Ratings (Remember/Know). These two had initially survived the audit but were dropped for consistency - adding a graded scale to old/new or R/K judgments is the same kind of add-on as the six already dropped.

5.6 Computerized vs. manual versions

Rule: A computerized version is a genuine variation only if it changes the sensory modality or motor activity the participant uses. Manipulating physical objects vs. clicking on a screen qualifies. Handwriting vs. key-pressing qualifies. Stylus-on-tablet vs. pencil-on-paper does not (same motor modality). Automated scoring or administration software alone is never a variation (that is MEAS or DESG).

Kept computerized/digital variations (6): Computerized (eCorsi) - physical block tapping -> screen tapping changes spatial/motor demands; Computerized DSST/SDMT - handwriting -> button pressing is a different fine motor skill; Computerized Mirror Tracing - loss of haptic edge feedback from physical template; Computerized Adaptive Version (Raven’s) - adaptive algorithm changes which items are presented, a procedural change beyond interface; Computerized vs. Physical Versions (Tower of London) - grasping/placing 3D beads -> drag-and-drop; PD against Computer Opponents (Prisoner’s Dilemma) - not an interface change but a structural one: deterministic program opponent vs. human.

Dropped (4): Computerized RAVLT (digital administration with automatic scoring - MEAS/DESG), Digital TMT/dTMT (stylus on tablet vs. pencil on paper - same motor modality; kinematic recording is MEAS), Computerized UFOV (UFOV is inherently screen-based; this is commercial software with automated staircase - DESG), Computerized WCST/WCST-CV (digital version with automated scoring - DESG).