Proactive Control of Episodic Memory Retrieval
[email protected]
Michael 1
1,2 Siena ,
Petar
1,3 Raykov ,
Maxime
1,4 Kraus
and Alexa
School of Psychology, Philsophy and Language Sciences, University of Edinburgh, 2 UK; Experimental Psychology, University of Cambridge, UK; 3 Psychology, University of Sussex, UK; 4 Cognitive Science Center, University of Amsterdam
RETRIEVAL ORIENTATION
AIM • Can we generalize associations between individual differences in cognitive control and goal-driven prioritization of recollection?
BACKGROUND • People can target memory strategically, e.g. recall events associated with a specific place, time or activity [1,2].. This control of recollection is thought to involve more general cognitive control abilities [1,3,4] • High working memory capacity (WMC) individuals may be more likely to control behaviour proactively [5], in line with goals, and to prioritize recollection according to goals [4,6-9]
• N = 36 right handed (7 male; 18 - 30 years; 1 exclusion). Controlled recollection task in 12 study-test cycles (Fig 1: 3 cycles).
WMC and LEFT PARIETAL OLD/NEW EFFECT
• Retrieval orientation: differential processing of retrieval cues (e.g. words in a memory test) according to the goal [2]. ERPs to unstudied (Novel) items. A priori analysis from prior Artist vs. Function study [11], diff significant in 500-700, 700-900 and 900-1100 ms time windows.
WMC (O-Span, partial) Target Old/new
WMC and ERP RETRIEVAL ORIENTATION Waveforms (FC6) All participants
4 mV
0 4 mV
Target Hits
501 ms - 601 ms
502 ms - 601 ms -2.65 µV
0 µV
2.65 µV
-1.89 µV
0
400
800
0 µV
1.89 µV
PRIORITIZATION OF RECOLLECTION
1200
WMC (O-Span, partial) Low WMC
4 mV
• At LP electrodes [P5, P3, P1] WMC negatively predicted Target-Nontarget ERP old/new effect, p = .0078, d = .47, N = 35 [robust regression; significant also with the 1 outlier with low O-Span).
899 ms - 1098 ms
902 ms - 1101 ms
900-1100 ms Pleasant-Artist Main effect -1.49 µV
0 µV
1.49 µV
900-1100 ms Artist-Function Varies with WMC -0.64 µV
0 µV
• Left parietal old/new effect: ERP correlate of recollection [2,12]. Predicted greater prioritization of Target vs. Nontarget recollection in high-WMC, i.e. positive association of LP effect with O-Span [6-9] • As in [9], L > R lateralized parietal old > new ERPs 500-600 ms, old > new p = .00005, h2g = .08; Targ > NT, p = .025, h2g = .02, N=36.
High WMC
0.64 µV
0
400
800
1200
SUMMARY
www.PosterPresentations.com
800
Novel CR
Pleasant Novel CR
• Consistent w/ ERPs, in F condition, Low O-Span slower for Nontargets only, ? prioritizing Target recollection unlike High O-Span (p = .048).
400
Nontarget CR
Function Novel CR
• Compare Targets vs. Nontarget performance by task, expect slower RTs for Nontargets if Target hits based on Target recollection only [10]. RTs slower for Targets vs. Nontargets only for P task (Condition x Task p = .021, h2g = .11; pairwise p < .001 for P, else p > .18)
Nontarget Old/new
ERP RO effect for Artist-Function
Artist Novel CR
• Target accuracy for Artist > Function (Pr: A = .81; F = .76, P = .79).
High
• Earlier time windows not significant unlike in [11].
• Standardised cognitive tests of WMC: Operation Span, Proactive vs reactive control: AX-CPT.
EXCLUSION TASK PERFORMANCE
ERP difference: Target-Nontarget Low
• Across 15 electrode pools (Frontal to Parietal, L/M/R), Pleasantness more positive than Artist 900-1100 ms, main effect, p = .046, h2g = .02.
• Test: recognition exclusion, vary retrieval goals each cycle, targeting one source (e.g. Respond “yes” only to words studied in the Artist task. Sources were the 3 different semantic encoding tasks.
• Test phase EEG from 64 electrodes, ext. 10-20, BioSemi ActiveTwo. Ref. ave. mastoid, 0.1–30 Hz band-pass,, ICA Ocular Correction in BVA 2.0
Waveforms (P3) (O-Span median split)
• Function ERPs more positive than Artist as in [11] but only in high WMC. At a priori [from 11] R Frontocentral electrodes [FC4 FC6 FT8] WMC positively predicted A-F ERP effect, p = .0054, d = .51.
METHODS
RESEARCH POSTER PRESENTATION DESIGN © 2015
1 Morcom
• Retrieval orientation ERP effect for Artist vs. Function task replicates [11] although weaker, later onsetting • High-WMC people engage retrieval orientation more strongly as predicted. Differs from [9], a null result for RO tho’ smaller N was part motivation for present study [unpubl data]. • High-WMC people also show LESS prioritization of recollection on ERP old/new measure, may reflect recall-to-reject as well as recall-to-accept; robust but not predicted, fits w/ RTs [10] • No significant associations with proactive attentional control (AX-CPT). WMC more direct index of required processes?
• On median split, Target > Nontarget old/new only differed in low performers, p = .004, d = .8, N=17 (for high, p = .244, N=18). • N.B. LP effect typically 500-800 ms; as in [9], overlapping negative-going old/new effect, Late Posterior Negativity (LPN) from 600-800 ms. LPN for Targ-NT also predicted by O-Span but weaker, p = .030, d = .37, N=35. No association of earlier old/new effects with O-Span.
SELECTED REFERENCES 1. Johnson, M.K. & Raye, C.L. (2000) In: Memory, brain & belief, Schacter, D.L. and Scarry, E., Eds, pp 35-86, Harvard UP. 2. Rugg, M. D. & Wilding, E. L. (2000). Trends Cogn Sci, 4, 108 – 115. 3. Morcom, A. M. (2016). Curr Dir Psychol Sci, 25, 143 – 150. 4. Unsworth, N., Spillers, G.J., Brewer, G.A. (2012). Memory, 20(2), 167-76. 5. Redick, T.S. (2014). Acta Psychol (Amst), 145(1), 1-9. 6. Elward, R. L. & Wilding, E. L. (2010). Brain Res, 1342, 1342 – 1355. 7. Sprondel, V., Kipp, K.H., & Mecklinger A. (2012). Dev Sci. 15(3), 330-44. 8. Elward, R. L., Evans, L.H., & Wilding, E. L. (2013). Cortex, 49(6), 1452-62. 9. Keating, J., Affleck-Brodie, C. & Morcom, A. M. (2017). PLoS ONE, 12(7). 10. Rosburg, T., & Mecklinger, A. (2016). Psychon Bull Rev [Epub March 2017]. 11. Dzulkifli, M. A., & Wilding, E. L. (2005). Neuropsych, 43, pp. 1152 – 1162. 12. Rugg, M.D. & Curran, T. (2007). Trends Cogn Sci 11, 251-257.