[ CogSci Summaries home | UP | email ]
http://www.jimdavies.org/summaries/

---

Schacter, D.L.(1992). Understanding Implicit Memory: A Cognitive Neuroscience Approach. American Psychologist, 47(4), 559-569.

@Article{Schacter1992,
author = {Daniel L. Schacter},
title = {Understanding Implicit Memory: A Cognitive Neuroscience Approach},
journal = {American Psychologist},
year = {1992},
volume = {47},
number = {4},
pages = {559--569}
}

Author of the summary: Michelle Sappong, 2011, naana.sappong@live.ca

Cite this paper for:

• Implicit memory: “unintentional non-conscious” [p559]
• Explicit memory: “intentional, conscious retrieval" [p559]
• dissociations are necessary conditions of the multiple systems view, but not sufficient [p560]
• evidence from cognitive neuroscience constrains processing views
• cognitive neuroscience aprroach to implicit memory encourages cross-domain hypothesis-generation and testing
• "type of processing is the crucial determinant of dissociations” (Roediger 1990 & Blaxton 1989) [p561]
• "perceptual (bottom-up) and conceptual (top-down) priming depend on different systems” [p.562]
• perceptual representation system (PRS) is independent of semantic memory system [p561]
• PRS plays a role on data-driven implicit priming [p561]
• transfer-appropriate processing: memory performance increased when encoding and retrieval processes use same operations [p.561]
• left hemisphere subsystem computes abstract word form representations [p562]
• right hemisphere subsystem computes specific word form representations [p562]
• When words are presented to the right hemisphere, changing a feature reduces priming. [p562]
• When words are presented to the left hemisphere, priming is the same. [p562]
• object decision priming is mediated by a PRS subsystem that computes global representations of objects. [p563]
• priming involves the visual word form system (as opposed to individual letter activation alone) [p564]
• auditory priming is little affected by semantic or pitch encoding [p566]
• left hemisphere processes phonemes; right hemisphere processes prosody [p566]
• right hemisphere PRS subsystem involved in voice specificity effects [p567]
• cognitive neuroscience approach should be used in combination with other approaches [p568]

Implicit memory: “unintentional non-conscious” , tested through stem completion, word identification tasks [p559]
Explicit memory: “intentional, conscious retrieval”, tested through recall, recognition tasks [p559]

Dissociations between implicit and explicit memory can be explained by different memory systems, or the use of different processes. [p559]

Systems-processes debate [p560]:

1) dissociations occur because implicit and explicit memory rely on two different memory systems (semantic and episodic, respectively), or
2)dissociations are a result of different processing operations (bottom-up and top-down) in the same system

Observed Dissociations:

• On the stem completion task, both amnesic and non-amnesic participants show “direct/repetition priming” [p560] (intact implicit memory) - given a three-letter stem, they are more likely to complete it with a word that was previously presented to them.
• Depth of processing affects explicit memory, but does not increase priming.
• Changing the modality of words at time of presentation and time of test reduces priming effects, but does not affect explicit memory.

Roediger(1990) & Blaxton(1989): “type of processing is the crucial determinant of dissociations” [p561]

Implicit stem completion tasks make use of “data-driven” (bottom-up) processes.
Explicit recall and recognition tasks make use of “conceptually driven” processes (top-down) [p561]

Dissociations between implicit tasks: word completion vs. answering general knowledge question
Dissociations between explicit tasks: recall vs. recognition

Are dissociations enough to postulate separate systems? They are a necessary condition, not sufficient – post-hoc postulation of separate systems for all observed dissociations creates chaos, therefore evidence is needed independent of dissociations. [p561]

This paper argues that a cognitive neuroscience approach can:

a) Provide independent evidence for multiple processes and systems
b) Increase external validity by cross-domain hypothesis-generation and testing
c) Motivate experiments and account for findings

Multiple Systems View:

Evidence from cognitive neuroscience supporting multiple systems and processes :

PET scans - visual word form and semantic processing occur in two separate brain regions.[p561]

Brain-damaged patients who lack access to semantic knowledge can still show intact access to structural knowledge of words.[p561]

There is a perceptual representation system (PRS) that is independent of the semantic system. [p561]

PRS: involved in priming on data-driven implicit tasks – explains why these tasks are unaffected by depth of processing (semantic system), but affected by change in modality. [p562]
Semantic system: involved in priming on conceptually driven tasks. [p562]

Roediger(1990) & Blaxton(1989): “transfer-appropriate processing”- If processing operations at time of study match processing operations during testing, memory performance increases. [p561] This is not always the case; processing views can account for this through cognitive neuroscience findings.

Processing View:

Cognitive neuroscience has predictive value, ability to constrain processing views.

Neuroscience findings can predict when transfer-appropriate processing will occur.

Example 1) Left hemisphere computes abstract word representations; right hemisphere computes specific word representations. [p562]

Marsolek, Kosslyn & Squire (1992): Word completion task – “priming was reduced by case changes [upper or lower] when stems were presented to the right hemisphere” [p562]

Cognitive neuroscience accurately predicts that specific priming will occur when words are presented to the right hemisphere, and abstract priming will occur when words are presented to the left hemisphere. [p562]

Example 2) Cells in the inferior temporal (IT) cortex of monkeys are responsible for computing global structure. IT cells not affected by changes in retinal size or left-right reflections (mirror image). [p563]

PRS system may operate like IT cells. [p563]

Object decision task: implicit, structurally possible and impossible objects shown to participants; they must decide whether or not they have seen them before. [p562]

Object decision task findings [p563]:

a) priming effect occurs when 3D structure is intentionally studied beforehand as opposed to 2D features
b) priming occurs only for possible objects
c) linking objects to prior semantic knowledge has no efffect on priming
d) priming exists in patients with memory disorders.

Therefore, object decision priming is mediated by a PRS subsystem that computes structural descriptions in the form of global representations. This system is presemantic. [p562]

Cognitive neuroscience accurately predicts that priming on perceptual implicit tasks will be unaffected by changes in size, or mirror images. [p563] (Processing view alone would not be able to predict this)

Cross-domain hypothesis-testing: evaluating ideas in domains other than those where the hypothesis was generated. [p564]
Implicit memory research can inform research on letter-by-letter readers, and vice-versa.

Case study: P.T. is a letter-by-letter reader who showed robust priming on identifying previously studied words presented for 500ms, but did not show priming for nonwords. This is evidence that priming involves the visual word form system (as opposed to individual letter activation alone), and that letter-by-letter readers may have an intact word form system, despite difficulties in reading whole words. [p564]

Cross-domain hypothesis-generation: use ideas from various fields to generate hypotheses about implicit memory [p565].
For example, use findings from cognitive neuroscience to generate hypotheses about auditory implicit memory.

Findings: Patients exhibit dissociations between form and meaning; they cannot understand spoken words, but they can repeat them and write them down (they still have access to word form reps). Some can, however, understand them in visual modality. [p565]

Hypotheses Generated:
1. PRS subsystem handles auditory word forms separately from semantic info, therefore implicit memory on an auditory task should be unaffected by semantic processing. [p565]
2. Voice specificity should affect priming when participants encode voice characteristics during the study task. [p566]

Experiment:
Test auditory implicit memory on a white noise recognition task. Half participants studied pitch of words beforehand, other half categorized the words before test. Voice (male/female) was either same or different. Priming indicatedby more accurate recognition of previously heard words. [p565]

Results:
Explicit memory is higher when words are categorized at time of study than when the pitch is analyzed.[p566]
Auditory priming is little affected by either semantic or pitch encoding at time of study.[p566]
Voice change did not significantly affect priming.[p566]

Cognitive neuroscience explains these results: The left hemisphere processes abstract auditory information (phonemes), whereas the right hemisphere processes prosodic features (voice info, pitch). [p566]

Ross (1981): Left-ear advantage for some voice information. [p566]
Van Lancker, Cummings, Kreiman, & Dobkin (1988): Right hemisphere lesions impair voice recognition.[p566]
Zaidel (1978): right hemisphere struggles to process voice information in background noise. [p566]
The white noise on the above task impaired the right hemisphere, thus masking the effects of voice specificity on priming, and further studies eliminating the white noise found that priming is indeed reduced by voice changes (Shachter & Church, in press). [p567]

All of this leads to a new hypothesis: a right hemisphere PRS subsystem is involved in voice specificity effects. [p567]

Limitations of a cognitive neuroscience approach :

• useful only where neurophysiological and neuropsychological data exists and is straightforward [p567]
• literature may not address certain hypotheses [p568]

Summary author's notes:

---

Back to the Cognitive Science Summaries homepage
Cognitive Science Summaries Webmaster:

JimDavies (jim@jimdavies.org)