Recall to reject and contextual discrimination: The influence of contextual distinctiveness on the control of recollection in exclusion tasks

1
Recall to reject and contextual discrimination
The influence of contextual distinctiveness on
the control of recollection in exclusion tasks

• Marianne de Chastelaine
• PhD Supervisors Mick Rugg Chris Brewin
• Institute of Cognitive Neuroscience
• University College London

Acknowledgements Wellcome Trust
2
The exclusion task
Study List 1 deep encoding non-targets Stu
dy List 2 shallow encoding
targets
Test phase - target old -
non-target new -
unstudied new
LAMP
KITE
HOSE
DOLL
LAMP
TENT
TREE
BEAR
KITE
3
Process dissociation procedure
Study List 1 deep-encoding Study List 2
shallow-encoding
LAMP
non-targets
HOSE
DOLL
TENT
targets
BEAR
KITE
4
Alternative proposal
Study List 1 deep-encoding Study List 2
deep-encoding
LAMP
non-targets
HOSE
DOLL
TENT
targets
BEAR
KITE
5
ERP correlate of recollection
Recollection a left-lateralised parietal
positivity
Data from Rugg et al., Nature, 1998, 392, 595-598
6
Herron Rugg (2003a) design
Study List 1 Exps 1 2 deep
non-targets Study List 2 Exp 1 deep
Exp 2 shallow
targets
Test phase - target old -
non-target new -
unstudied new
LAMP
KITE
HOSE
DOLL
LAMP
TENT
TREE
BEAR
KITE
7
Herron Rugg (2003a) ERP data
Poor target memory
Good target memory

NTARG
TARG
NEW
5µV
0
600 msec
0
600 msec

• Good target memory target recognition based on
  information diagnostic of study source
  non-target rejection based on absence of this
  information.
• Poor target memory relatively little source
  information available for targets recollection
  of non-target source necessary for their
  rejection.
• Findings suggest the adoption of a retrieval
  strategy.
• retrieval orientation enables test cues to be
  processed in a way that selectively probes for
  target recollection.

8
Effect of study task similarity (1)

• In contrast to these findings, left parietal
  old/new effects for correctly rejected
  non-targets have been reported, despite good
  target accuracy (e.g., Cycowicz, Friedman
  Snodgrass, 2001 Cycowicz, Friedman Duff, 2003
  Wilding Rugg, 1997 Wilding Sharpe, 2004).
• Why do these findings differ?
• One difference between these ERP studies concerns
  the degree of similarity between target and
  non-target study contexts.
• Presumably, the adoption of a retrieval
  orientation, allowing exclusive recollection of
  targets, will be more successful when there are
  fewer overlapping contextual features associated
  with targets and non-targets engendered during
  the study phase.
• One factor that should modulate non-target
  recollection is the degree of similarity between
  target and non-target study contexts (cf. Wilding
  Sharpe, 2004).

9
Aims of current study

• To investigate whether the ERP correlates of
  recollection would differ according to the degree
  of similarity between target and non-target study
  contexts
• For the similar group, target and non-target
  study tasks were identical and, for the
  different group, target and non-target study
  tasks were more distinct.
• Predictions
• Similar group Given greater similarity between
  target and non-target study contexts, attempts to
  retrieve target source will, to some extent, give
  rise to the recollection of non-target source
  correctly rejected non-targets will elicit a left
  parietal old/new ERP effect.
• Different group With greater differences
  between target and non-target study contexts, a
  retrieval orientation allowing exclusive
  recollection of target source will be more
  successfully adopted non-targets will be
  rejected on the basis of the absence of this
  information and, thus, will fail to elicit a left
  parietal old/new ERP effect.

10
Design





Study list 1 non-targets
Study list 2 - targets
Test phase
Both groups
LAMP
60 targets old
- 60 items
- 60 items
KITE
60 non-targets new
Task 1 Both groups colour association
Task 1 Both groups colour association
TREE
Task 2 Similar group pleasantness rating
Different group indoor/outdoor task
Task 2 Both groups pleasantness rating
60 unstudied new

• Retaining the above correspondence of group and
  item type to study task, word colour and tasks
  were fully counterbalanced across study list and
  group.
• At test, the requirement was to respond with one
  key press to targets and with another key press
  to non-targets and unstudied items.

11
ERP recording

• Test phase only
• EEG recorded from 29 silver/silver chloride
  electrodes and referenced to linked mastoids
• Blink corrected using linear regression
• Sampling rate 125 Hz for 2048 ms (pre-stimulus
  baseline 104 ms)
• N16 (each group)

12
Behavioural data
Accuracy
RT

• There were no differences in accuracy or RTs as
  a function of group.
• Accuracy for new items was higher than for
  targets and non-targets. Accuracy for targets and
  non-targets did not differ.
• RTs for new items were faster than for targets
  and non-targets. RTs for targets and non-targets
  did not differ.

13
Non-target left parietal ERP effects
Similar group
Different group
0
800 ms
NT New 500-800 ms
NT New 300-500 ms

• For the similar group only, correctly rejected
  non-targets elicited a left-parietal old/new
  effect (red bar) which onset around 200-300 msec
  (blue bar).

14
Target left parietal ERP effects
Different group
Similar group

Target
New
0
800 ms
0
800 ms
T New 500-800 ms
T New 300-500 ms
T New 500-800 ms
T New 300-500 ms

• For both groups, left parietal (500-800 msec)
  old/new effects were elicited by targets. The
  same items also elicited early-onsetting (300-500
  msec) left parietal effects.

15
Interim discussion

• As predicted, for the similar group only,
  correctly rejected non-targets elicited a left
  parietal old/new effect. In this case, as there
  was greater overlapping contextual information
  between targets and non-targets engendered during
  the study phase, attempts to retrieve target
  source gave rise to the recollection of
  non-target source.
• For the different group, non-targets failed to
  elicit a left parietal old/new effect. As target
  and non-target study contexts were partially
  distinct, here a retrieval orientation, allowing
  exclusive recollection of target source, was more
  successfully adopted.
• However, in contrast to the above
  oberservations, there is some reason to believe
  that, for the different group, non-targets may
  well have been recollected even though there was
  no reliable left parietal effect elicited by
  these items.

16
Non-target left parietal ERP effects
Similar group
Different group
0
800 ms
NT New 500-800 ms
NT New 300-500 ms

• For the similar group only, correctly rejected
  non-targets elicited a left-parietal old/new
  effect (red bar) which onset around 200-300 msec
  (blue bar).

17
Late posterior negativity
Similar group
Different group

Target
Non-target
0
800 ms
0
800 ms
New
5µV
18
Late negativity - functional significance?

• A combination of processes that are both
  response-locked and stimulus-locked.
• Often reported in ERP studies in which source
  judgements have been required.
• Stimulus-locked component suggested to reflect
  processes related to the search for and/or
  maintenance of the conjunction of item and
  associated contextual information (Johansson
  Mecklinger, 2003).

19
Late posterior negativity
Similar group
Different group

Target
Non-target
0
800 ms
0
800 ms
New
5µV
20
Specified retrieval search



Study list 1 non-targets
Study list 2 - targets
Task 1 Both groups colour association
Task 1 Both groups colour association
Task 2 Similar group pleasantness rating
Different group pleasantness rating
Task 2 Similar group pleasantness rating
Different group indoor/outdoor task
21
Mis-matched contextual details

• Similar group
• Targets contextual details of 2nd study task
• Non-targets contextual details of 2nd study
  task

22
Specified retrieval search



Study list 1 non-targets
Study list 2 - targets
Task 1 Both groups colour association
Task 1 Both groups colour association
Task 2 Similar group pleasantness rating
Different group pleasantness rating
Task 2 Similar group pleasantness rating
Different group indoor/outdoor task
23
Mis-matched contextual details

• Similar group
• Targets contextual details associated with the
  2nd study task
• Non-targets contextual details associated with
  the 2nd study task
• Different group
• Targets contextual details associated with the
  visualisation task
• Non-targets details of cognitive operations
  associated with the visualisation task AND the
  2nd study task allocated to non-targets
• Processes acting upon a mis-match between the
  target memory representation and retrieved
  irrelevant memories likely to have been engaged
  to the greatest extent for non-targets from the
  different group.
• Is this reflected in the greater negativity
  elicited by these items?

24
Final conclusions

• There is some evidence that the adoption of a
  specific retrieval orientation can account for
  the finding that test words can be used to
  selectively retrieve episodic information
  involving target words as opposed to non-target
  pictures (Herron Rugg, 2003b).
• However, data from the present experiment
  suggest that even a partial overlap between
  target and non-target study contexts will lead to
  the failure of the adopted retrieval orientation
  to focus retrieval operations exclusively on
  target context.
• It seems that additional mechanisms would be
  required to assess and possibly act upon
  mismatches between the targeted memory
  representation and the contextual details that
  were actually retrieved it is proposed that the
  late posterior negativity reflects such
  processes.
• Whether such processes tentatively proposed to
  be reflected by the late negativity act to
  suppress irrelevant information or to simply
  register a mismatch, it would seem that these may
  act in concert with the adoption of a specific
  retrieval orientation to bias initial retrieval,
  and subsequent attention, towards relevant
  memories amongst competing alternatives.