Implicit working memory

Implicit working memory Hassin Bargh Engell McCulloch

Tag layer of the DHSS Meta-Tagging Project. The article text is reproduced from the staged copy named in sourceDesc and is not altered; the tags are held separately in standOff so that overlapping and nested spans are preserved.

Implicit working memory 2009 Consciousness and cognition 10.1016/j.concog.2009.04.003 Hassin2009-implicit-working-memory Hassin2009-implicit-working-memory.txt, the project's staged copy 2426d910b7fcc13dea8e96be4e832637c9727e12 not-redistributable cognitive-psychology

the paper's own words

the tag layer; a sentence written by the tagger about the paper, always anchored to a quotation but not itself a quotation

0 spans located by searching the stored text for the quote's exact words, allowing any whitespace between them whether the active components of working memory can operate without intention and outside conscious awareness Quotation withheld because this record is not cleared for redistribution. can working memory operate implicitly, given that every existing WM task presupposes that it cannot Quotation withheld because this record is not cleared for redistribution. shows across five studies that patterns are extracted, maintained and used to bias responding without intention or awareness, and argues the field's view of working memory must widen to include an implicit mode Quotation withheld because this record is not cleared for redistribution. existence Quotation withheld because this record is not cleared for redistribution. implicit working memory Quotation withheld because this record is not cleared for redistribution. behavioural-experiment behavioural-experiment Quotation withheld because this record is not cleared for redistribution. mechanisms Incidental extraction of the pattern generates an anticipation of the next disk's location, which speeds responses when the anticipation is correct and slows them when it is violated Quotation withheld because this record is not cleared for redistribution. mechanisms Conscious, intentional search for patterns produces frequent failures on the pattern-less majority of sets, and those failures interfere with the implicit extraction process Quotation withheld because this record is not cleared for redistribution. mechanisms Discovering a pattern itself imposes a brief processing cost - an implicit 'aha' - that partly cancels the speedup it produces Quotation withheld because this record is not cleared for redistribution. the field's instruments are read as evidence about the field: every standard WM task is explicit, so the tasks record what researchers believe rather than test it Quotation withheld because this record is not cleared for redistribution. the question cannot be asked with the existing instruments because they presuppose its answer, so the paradigm has to be built before the experiment can exist Quotation withheld because this record is not cleared for redistribution. a functional definition is adopted deliberately, so that whether the new task counts as a working memory task is settled by what it requires rather than by precedent Quotation withheld because this record is not cleared for redistribution. the design forecloses statistical learning by construction: pattern-following and pattern-breaking endings are equiprobable, so learning history predicts chance on the critical disk Quotation withheld because this record is not cleared for redistribution. the first block is analysed separately as the empirical version of the same argument, since each pattern has appeared only once and there is nothing yet to have learned Quotation withheld because this record is not cleared for redistribution. awareness is attacked from four directions rather than one — open debriefing, immediate reconstruction, forced-choice recognition, and a correlation between reconstructed and actual moves Quotation withheld because this record is not cleared for redistribution. in the fourth study awareness is manipulated instead of measured, which converts a null result about awareness into a predicted dissociation Quotation withheld because this record is not cleared for redistribution. the excluded participants are turned into a test: if awareness drove the effect it should be larger in those who noticed the patterns, and it is smaller Quotation withheld because this record is not cleared for redistribution. a self-undermining fact — that every finite sequence follows some rule — is stated and then shown to bias the study against its own hypothesis Quotation withheld because this record is not cleared for redistribution. the alternative account is not tested but reasoned through: imagine long-term memory doing this task, and it turns out to need the very operations attributed to working memory Quotation withheld because this record is not cleared for redistribution. stimulus position is jittered specifically to block a strategy that would let attention narrow to the local level and bypass the global stimulus Quotation withheld because this record is not cleared for redistribution. the fifth study changes the material from spatial to algebraic so the claim rests on the function rather than on the modality Quotation withheld because this record is not cleared for redistribution. Whether pattern naming, anticipation and control are done by working memory alone or by working memory interacting with long-term memory is left unresolved. what this limits: the attribution to working memory alone: WHETHER THE PROCESSES OF NAMING THE PATTERNS, CREATING ANTICIPATIONS FOR FUTURE LOCATIONS, AND CONTROLLING COGNITION, ARE DONE SOLELY BY WM or involve other systems as in other WM paradigms Quotation withheld because this record is not cleared for redistribution. The comparison between excluded (aware) and included participants is heuristic rather than a formal test. Quotation withheld because this record is not cleared for redistribution. The unequal group sizes prevented a direct statistical comparison of aware and unaware participants. what this limits: the aware-versus-unaware comparison: GIVEN THESE HUGE DIFFERENCES IN NUMBERS WE CANNOT DIRECTLY COMPARE THE EFFECTS USING AN ANOVA Quotation withheld because this record is not cleared for redistribution. The proposed implicit "aha" account of the asymmetry between speedup and slowdown is speculative. Quotation withheld because this record is not cleared for redistribution. the subject pool was contaminated by the laboratory's own research programme — participants arrived suspecting a hidden manipulation what this limits: every unawareness claim in the paper: HASSIN'S LAB (WHERE THIS STUDY WAS CONDUCTED) RUNS MANY NON-CONSCIOUS GOAL PRIMING STUDIES, and TOWARDS THE END OF THE ACADEMIC YEAR, MANY PARTICIPANTS COME TO THE LAB SUSPECTING THAT THERE IS MORE TO OUR EXPERIMENTS Quotation withheld because this record is not cleared for redistribution. the excluded-participant analysis is conceded to be informal, compared by effect size because the group sizes forbid a test what this limits: the same scope as limitations[2], with what was done instead: HENCE, WE USE AS HEURISTICS THE ACTUAL SIZE OF THE EFFECT IN MILLISECONDS AND EFFECT SIZES Quotation withheld because this record is not cleared for redistribution. the predicted pattern is asymmetric in a way the hypothesis did not anticipate: the slowdown is reliable and the speedup mostly is not what this limits: which comparison held in which study: THE DIFFERENCES BETWEEN BROKEN PATTERN AND CONTROL SETS WERE SIGNIFICANT IN THREE STUDIES (1, 2, & 3), AND MARGINALLY SIGNIFICANT IN ONE (4), while the Pattern-versus-control difference was significant in fewer Quotation withheld because this record is not cleared for redistribution. the explanation offered for that asymmetry is labelled speculation by the authors Quotation withheld because this record is not cleared for redistribution. establishing that anything is non-conscious is acknowledged as never fully achievable Quotation withheld because this record is not cleared for redistribution. whether long-term memory also contributes is left open, defended by parsimony rather than by data duplicate of: limitations[0] · what this limits: the same scope as limitations[0] - whether long-term memory also contributes, left open Quotation withheld because this record is not cleared for redistribution. block-level data exist for only two of the five studies because the analysis was thought of late what this limits: block-level analyses, and only two studies have them: THE IDEA TO EXAMINE THE EFFECTS OF PATTERNS IN THE FIRST BLOCK CAME PRETTY LATE IN THE PROJECT. HENCE, WE HAVE INFORMATION ABOUT BLOCKS ONLY IN EXPERIMENTS 1 AND 4 Quotation withheld because this record is not cleared for redistribution. whether implicit and explicit WM are one system in two modes or two systems is unresolved what this limits: the one-system-or-two question: WHETHER THEY BELONG TO THE SAME SYSTEM OR TO TWO SEPARATE SYSTEMS, and what the PUTATIVE DIFFERENCES BETWEEN EXPLICIT AND IMPLICIT WM are Quotation withheld because this record is not cleared for redistribution. the compound stimulus — a large digit built from small characters — borrowed so the cover task can be about the components while the pattern lives in the whole Quotation withheld because this record is not cleared for redistribution. the finding that conscious search damages extraction of complex structure, imported as the prediction that an explicit version of the task will show no effect Quotation withheld because this record is not cleared for redistribution. a YOKED control condition in which the first three disk locations are random and only the fourth and fifth follow the pattern - the mirror image of the experimental condition, matched on everything except where the regularity sits eliminates: that performance on the fourth and fifth disks reflects the regularity carried by the first three - the yoked condition matches everything except WHICH disks carry the pattern · outcome (stated by the tagger, no quotation): absent-stated-about-other · would have ruled out, had it been run: the mere presence of a pattern, and the visual properties of the display, as the explanation - both conditions contain a pattern, so a difference between them cannot be attributed to having one Quotation withheld because this record is not cleared for redistribution. working memory — defined functionally, by the operations a task demands, rather than by membership in the established task set; the move that makes the new paradigm admissible Quotation withheld because this record is not cleared for redistribution. working memory — as the field's tasks define it in practice: materials explicitly presented and explicitly manipulated Quotation withheld because this record is not cleared for redistribution. implicit — as unintentional (not deliberately recruited) versus as unconscious (not available to awareness); the paper claims both and tests them separately Quotation withheld because this record is not cleared for redistribution. conscious involvement in WM — as attaching to some components (Cowan, Baddeley) versus to all active operations (Baars and Franklin) Quotation withheld because this record is not cleared for redistribution. pattern — visual, drawn by connecting disk locations, versus conceptual, constituted by the algebraic relation between successive numbers Quotation withheld because this record is not cleared for redistribution. pattern — the one the experimenters built in, versus the mathematical rule any finite sequence trivially satisfies; the second sense makes every broken pattern a pattern Quotation withheld because this record is not cleared for redistribution. implicit working memory — as a mode of one system versus as a separate system; the paper says which it assumes and why, and leaves the question open Quotation withheld because this record is not cleared for redistribution. extraction — deliberately left open to cover identification and recognition, because the argument needs only the operations, not the label Quotation withheld because this record is not cleared for redistribution. claim_type: existence a second reading reaching the same value. Recorded: existence — note: a second reading reaching the same value from a different sentence. Not applied - a person decides. Quotation withheld because this record is not cleared for redistribution. method: two purpose-built incidental paradigms — sequential disks forming spatial patterns, and sequential digits forming algebraic ones — with a perceptual cover task and four converging awareness checks a second reading DIFFERING from the recorded value. Recorded: behavioural-experiment — note: a second reading, verbatim-verified, disagreeing with the value on the record. Not applied - a person decides. Quotation withheld because this record is not cleared for redistribution. phenomenon: implicit working memory — unintentional, non-conscious operation of the active components of working memory a second reading DIFFERING from the recorded value. Recorded: implicit working memory — note: a second reading, verbatim-verified, disagreeing with the value on the record. Not applied - a person decides. Quotation withheld because this record is not cleared for redistribution. replicates → McKone (1995) verbal and Maljkovic & Nakayama (1994) visual implicit short-term memory findings Quotation withheld because this record is not cleared for redistribution. extends → previous implicit short-term memory findings (McKone; Maljkovic and colleagues) Quotation withheld because this record is not cleared for redistribution. critiques → the prevalent view that the active components of working memory are intentional and conscious, and the task set that encodes it Quotation withheld because this record is not cleared for redistribution. contradicts → Baars and Franklin's claim that all active components of classical working memory are conscious Quotation withheld because this record is not cleared for redistribution. extends → McKone's and Maljkovic's demonstrations of implicit short-term memory, from maintenance alone to updating, computation and control Quotation withheld because this record is not cleared for redistribution. supports → the social-cognition literature showing that high-order processes and goal pursuit can occur non-consciously Quotation withheld because this record is not cleared for redistribution. extends → models of non-conscious high-order processes, which currently contain no working memory component Quotation withheld because this record is not cleared for redistribution. builds-on → the implicit-learning finding that directing attention to a complex structure can prevent its extraction Quotation withheld because this record is not cleared for redistribution. implicit working memory —part-of→ Working Memory implicit working memory Working Memory strength of the relation: 0.9 Quotation withheld because this record is not cleared for redistribution. implicit working memory —distinct-from→ implicit learning implicit working memory implicit learning strength of the relation: 0.95 Audit 2026-09-28: DOUBTFUL - read against its full span and kept, but the span writes 'pattern extraction', not 'implicit working memory'. Quotation withheld because this record is not cleared for redistribution. implicit working memory —distinct-from→ long term memory implicit working memory long term memory strength of the relation: 0.85 Audit 2026-09-28: DOUBTFUL - read against its full span and kept, but the span POSES the question whether the effect is working memory or long-term memory; distinct-from is not asserted as a finding. Quotation withheld because this record is not cleared for redistribution. Pattern sets —contrasted-with→ Broken Pattern Pattern sets Broken Pattern strength of the relation: 1 Audit 2026-09-28: DOUBTFUL - read against its full span and kept, but the span writes 'Broken Pattern condition' and 'a set'; the Pattern sets are named in the previous sentences. Quotation withheld because this record is not cleared for redistribution. pattern extraction —operationalised-as→ reaction time pattern extraction reaction time strength of the relation: 0.95 Audit 2026-09-28: DOUBTFUL - read against its full span and kept, but the span names reaction time as the dependent measure; 'pattern extraction' is not in it, so operationalised-as is loose. Quotation withheld because this record is not cleared for redistribution. pattern extraction —causes→ anticipations pattern extraction anticipations strength of the relation: 0.9 Quotation withheld because this record is not cleared for redistribution. conscious awareness —operationalised-as→ immediate reconstruction conscious awareness immediate reconstruction strength of the relation: 0.95 Audit 2026-09-28: DOUBTFUL - read against its full span and kept, but the span writes 'awareness' and 'reconstruct', not 'conscious awareness' / 'immediate reconstruction'. Quotation withheld because this record is not cleared for redistribution. conscious awareness —operationalised-as→ recognition test conscious awareness recognition test strength of the relation: 0.9 Audit 2026-09-28: DOUBTFUL - read against its full span and kept, but a recognition test was GIVEN after each set; operationalised-as is loose. Quotation withheld because this record is not cleared for redistribution. Implicit condition —contrasted-with→ Explicit condition Implicit condition Explicit condition strength of the relation: 1 Quotation withheld because this record is not cleared for redistribution. ANOVA ANOVA Quotation withheld because this record is not cleared for redistribution. repeated-measures ANOVA repeated-measures ANOVA Quotation withheld because this record is not cleared for redistribution. t-test (planned comparisons) t-test (planned comparisons) Quotation withheld because this record is not cleared for redistribution. effect size effect size Quotation withheld because this record is not cleared for redistribution. correlation correlation Quotation withheld because this record is not cleared for redistribution. standard deviation (outlier trimming) standard deviation (outlier trimming) Quotation withheld because this record is not cleared for redistribution. interaction test in a mixed factorial design interaction test in a mixed factorial design Quotation withheld because this record is not cleared for redistribution. partial η 2 partial η 2 Quotation withheld because this record is not cleared for redistribution. within subjects’ ANOVA within subjects’ ANOVA Quotation withheld because this record is not cleared for redistribution. Planned comparisons Planned comparisons Quotation withheld because this record is not cleared for redistribution. MSE MSE Quotation withheld because this record is not cleared for redistribution. mixed design mixed design Quotation withheld because this record is not cleared for redistribution. one tailed one tailed Quotation withheld because this record is not cleared for redistribution. correlation correlation Quotation withheld because this record is not cleared for redistribution. Participants responded faster to the final disk of pattern-following sets than to pattern-breaking sets, with control sets in between, even though the patterns were irrelevant to their explicit task. Quotation withheld because this record is not cleared for redistribution. The pattern effect was already present in the very first block, which rules out gradual implicit learning as its source. Quotation withheld because this record is not cleared for redistribution. Judges could detect no awareness of the patterns in participants' immediate reconstructions of the last set they had seen. Quotation withheld because this record is not cleared for redistribution. Only 5 of the remaining 30 participants (20%) correctly recognized the last set. Quotation withheld because this record is not cleared for redistribution. The pattern effect appeared only when instructions kept the task incidental; telling participants to search for patterns abolished it. Quotation withheld because this record is not cleared for redistribution. The effect replicated conceptually with numeric (algebraic) rather than visuospatial patterns. Quotation withheld because this record is not cleared for redistribution. The effect is asymmetric: the slowdown on broken-pattern sets is more reliable than the speedup on pattern sets. Quotation withheld because this record is not cleared for redistribution. Participants excluded for reporting awareness showed a smaller, not larger, pattern effect than the unaware participants. Quotation withheld because this record is not cleared for redistribution. An accompanying fMRI analysis found right prefrontal activation for pattern relative to control sets. Quotation withheld because this record is not cleared for redistribution. responses to the last disk were faster after an intact pattern than after a broken one, with controls in between Quotation withheld because this record is not cleared for redistribution. the pattern effect was already highly significant in the first block Quotation withheld because this record is not cleared for redistribution. reconstruction correlations with the actual moves did not differ from zero, individually or on average Quotation withheld because this record is not cleared for redistribution. recognition of the last set was 5 of 30 (20%), exactly the chance rate Quotation withheld because this record is not cleared for redistribution. the effect held even when only participants who failed the recognition test were analysed Quotation withheld because this record is not cleared for redistribution. telling participants the patterns exist abolished the effect — a significant Instructions by Pattern interaction, with the effect present implicitly and absent explicitly Quotation withheld because this record is not cleared for redistribution. the effect replicated with algebraic patterns: fourth terms were answered faster in pattern than in broken-pattern sets Quotation withheld because this record is not cleared for redistribution. the participants excluded for being aware showed a SMALLER effect than those included — 17 ms against 31 ms, with effect size half again as large in the unaware group Quotation withheld because this record is not cleared for redistribution. there is both speedup and slowdown, with the slowdown after a broken pattern the stronger Quotation withheld because this record is not cleared for redistribution. the companion fMRI work found increased right prefrontal activation during pattern sets relative to controls Quotation withheld because this record is not cleared for redistribution. prior work found an implicit short-term memory component in visual popout lasting 15-30 s Quotation withheld because this record is not cleared for redistribution. in study 2 the pattern effect replicated with faster overall responding, 462 ms against 489 ms for broken patterns Quotation withheld because this record is not cleared for redistribution. working memory paradigm working memory paradigm Quotation withheld because this record is not cleared for redistribution. subset hypothesis subset hypothesis Quotation withheld because this record is not cleared for redistribution. active contents or processes active contents or processes Quotation withheld because this record is not cleared for redistribution. Pattern sets condition Pattern sets condition Quotation withheld because this record is not cleared for redistribution. implicit working memory implicit working memory Quotation withheld because this record is not cleared for redistribution. subset hypothesis subset hypothesis Quotation withheld because this record is not cleared for redistribution. a working memory paradigm a working memory paradigm Quotation withheld because this record is not cleared for redistribution. Pattern sets Pattern sets Quotation withheld because this record is not cleared for redistribution. Broken Pattern Broken Pattern Quotation withheld because this record is not cleared for redistribution. Control Control Quotation withheld because this record is not cleared for redistribution. lag 0 lag 0 Quotation withheld because this record is not cleared for redistribution. Random Random Quotation withheld because this record is not cleared for redistribution. active contents or processes active contents or processes Quotation withheld because this record is not cleared for redistribution. reaction time reaction time Quotation withheld because this record is not cleared for redistribution. instructions instructions Quotation withheld because this record is not cleared for redistribution. reaction time to the last disks reaction time to the last disks Quotation withheld because this record is not cleared for redistribution. Pattern: Pattern, Broken Pattern, Control Pattern: Pattern, Broken Pattern, Control Quotation withheld because this record is not cleared for redistribution. Instructions: Implicit vs. Explicit Instructions: Implicit vs. Explicit Quotation withheld because this record is not cleared for redistribution. Block Block Quotation withheld because this record is not cleared for redistribution. awareness awareness Quotation withheld because this record is not cleared for redistribution. lag lag Quotation withheld because this record is not cleared for redistribution. M = 528 ms vs. M = 565 ms — mean reaction time to the last disk in Pattern versus Broken Pattern sets in Study 1 Quotation withheld because this record is not cleared for redistribution. 31 ms / 17 ms — size of the broken-pattern minus pattern RT difference for included (unaware) versus excluded (aware) participants Quotation withheld because this record is not cleared for redistribution. 20% — proportion of participants who correctly recognized the last pattern set, exactly the chance rate in a 5-alternative test Quotation withheld because this record is not cleared for redistribution. 24 (columns) by 18 (rows) — dimensions of the matrix into which the display screen was divided for disk locations Quotation withheld because this record is not cleared for redistribution. Twenty — number of participants in Study 1 Quotation withheld because this record is not cleared for redistribution. partial η 2 = .40 — effect size of the Pattern factor within the Implicit instruction condition in Study 4 Quotation withheld because this record is not cleared for redistribution. M = 547 ms, SD = 84 vs. M = 561 ms, SD = 82 — mean RT to the fourth term in Pattern versus Broken Pattern number sets in Study 5 Quotation withheld because this record is not cleared for redistribution. 24 by 18 — the matrix of columns and rows dividing the screen Quotation withheld because this record is not cleared for redistribution. five millimeters — disk diameter Quotation withheld because this record is not cleared for redistribution. sets of five — stimuli per set, demarcated by a 1500 ms fixation point Quotation withheld because this record is not cleared for redistribution. 150 ms — interval between response and the next disk Quotation withheld because this record is not cleared for redistribution. 10 patterns, 10 broken, 10 controls, 70 random — composition of each 100-set block Quotation withheld because this record is not cleared for redistribution. 3.125 — average distance in matrix units between successive disks in experimental sets Quotation withheld because this record is not cleared for redistribution. 528 vs 565 ms — Study 1 pattern against broken pattern reaction times Quotation withheld because this record is not cleared for redistribution. 20% — recognition accuracy in the five-alternative test, exactly chance Quotation withheld because this record is not cleared for redistribution. 31 ms vs 17 ms — the pattern effect in included (unaware) against excluded (aware) participants Quotation withheld because this record is not cleared for redistribution. partial η 2 = .20 and .14 — effect sizes for included against excluded participants Quotation withheld because this record is not cleared for redistribution. 20 excluded, 79 included — participants across studies 1–4 in the awareness comparison Quotation withheld because this record is not cleared for redistribution. 15–30 s — duration of Maljkovic's implicit popout memory effect Quotation withheld because this record is not cleared for redistribution. ∼140 ms at lag 0, asymptoting at approximately 50 ms at lag 4 — McKone's short-term repetition priming component Quotation withheld because this record is not cleared for redistribution. 200 ms / three SDs — the response-time exclusion window applied in every study Quotation withheld because this record is not cleared for redistribution. twenty — participants in Study 1 Quotation withheld because this record is not cleared for redistribution. fifty-four — participants in Study 4, the implicit/explicit contrast Quotation withheld because this record is not cleared for redistribution. twenty-eight — participants in Study 5, the algebraic replication Quotation withheld because this record is not cleared for redistribution. 25% of the trials — share of pattern and broken-pattern sets in Study 5 Quotation withheld because this record is not cleared for redistribution.

The full text of this article is not redistributable, so it is not included here. The evidence sentences are withheld with it; the tag layer and its grounding status remain. The article itself is at https://doi.org/10.1016/j.concog.2009.04.003.