Chapter 9: Recognition

advertisement
Benjamin Allred
벤자민 알레드
HUMAN MEMORY
CHAPTER 9: RECOGNITION
Contents
 Questions to Think About
 Definitions
 Recognition Versus Recall
 Single Process Models
 Generate-Recognize Models
 Remember Versus Know
 The Mirror Effect
 Face Recognition
Questions to Think About
 Which test is easier – a recognition test or a
recall test? (What makes one test easier than
another?)
 How are remember and know judgments
related to explicit and implicit memory?
 Why is it easier to recognize faces of one’s
own age-group?
Definitions
 Recall
 Recognition
 Distractors/lures
 In a recall test, the experimenter provides the
context and the subject has to retrieve the
target; in a recognition test, the experimenter
provides the target and the subject has to
retrieve the context. (Hollingworth (1913))
Recognition Versus Recall
 Recognition experiment - Shepard (1967)
 Subjects presented with lists of stimuli
 Words, sentences, photographs
 At test, presented with two stimuli, one from
original list, one new
 Words: 88%
 Sentences: 89%
 Pictures: almost 100%
Recognition Versus Recall
 Recall experiment – Mäntylä (1986)
 Subjects presented with lists of words, for which
they had to generate three properties for each
 At test, experimenter presented the properties
 Subjects recalled approximately 91% of the words
Types of Recognition Tests
 Alternative Forced Choice (2AFC, 4AFC)
 Given multiple choices, choose the one already
seen
 Yes-No
 Given one choice, indicate whether the item is
“old” or “new”
Yes-No Recognition Test
Possible Outcomes in a Yes-No Recognition Test
Subject’s Response
Test Item
Yes
No
Old
Hit
Miss
New
False Alarm
Correct Rejection
Single Process Models
 Early theories of recognition
 Tagging Model
 When an item occurs, it is tagged with the relative time
of occurrence
 Strength Theory
 The more recent the item, the stronger or more familiar
it is
 Limitations
 These models contain only a single process
 (Meaning that the same manipulation (word frequency,
intentionality, etc) should have the same effect on both
recall and recognition)
Evidence of Limitations
Generate-Recognize Models
 Two-stage models
 Recall is made up of two processes
 First, generate a set of plausible candidates for recall
(generation stage)
 Second, confirm whether each word is worthy of
being recalled (recognition stage – not the same as
the recognition test)
 Recognition is made up of only one process
 Because the experimenter provides a candidate,
recognition does not need the generation stage
Generate-Recognize Models
 Example: HAM (human associative memory)
(Anderson and Bower (1973))
 Assumes words are stored in associative network
 As words are presented, they are tagged with a
contextual marker
 Pathways to associated words are also tagged
 At recall:
 Contextual markers are followed to generate a set of
plausible candidates (generation stage)
 After examining number of associations between target
word and context, “old” or “new” is chosen depending on
sufficient contextual evidence (recognition stage)
Generate-Recognize Models
 Solves limitations of single process model
 The same manipulation does not have to have the
same effect on both recall and recognition
 Have problems of their own, however
 They require that if a word can be recalled, it must also
be recognized
 Because the second stage is common to both recall and
recognition, a successful outcome in one test should
mean a successful outcome for the other
 Recall failure is quite common and explainable, but
recognition failure is contrary to the prediction of
generate-recognize models
Recognition Failure
 Experiment by Watkins & Tulving (1975)
 Proved that a word could be recalled, even though
it could not be recognized
Step
Procedure
Example
1a
1b
List 1 presented
Cued recall of List 1
badge-button
badge-button
2a
2b
List 2 presented
Cued recall of List 2
preach-rant
preach-rant
3
List 3 presented
glue-chair
4a
4b
Free association stimuli presented
Free association responses made
table
table-chair, cloth, desk, dinner
5a
5b
Recognition test sheets presented
Recognized items circled
desk top chair
desk top chair
6
Cued recall of List 3
glue-chair
Generate-Recognize Models
 Adding a search process during recognition
stage could allow a generate-recognize
model to account for recognition failure
 Familiarity instantly computed to make response
 If familiarity value is not decisive enough, a search
is performed
 In the previous experiment, the target word (chair) is
not “found” in the search because the retrieval
phase (step 5) contained inappropriate cues
 The recall test (step 6) provided appropriate cues, so
the search process is successful
Remember Versus Know
 Relatively recent change in recognition
methodology (1985, 1988)
 Does someone specifically remember or just
somehow know?
 Experiment (Tulving (1985)):
 Present subjects with category-member pairs
 Recall tests:
 Free recall test
 Cued recall test (category)
 Cued recall test (category + first letter of target)
 The proportion of remember judgments decreased
over the three kinds of tests
Remember Versus Know
 Gardiner (1990, 1993) gives an explanation:
 Remember judgments are influenced by conceptual
and attentional factors
 Know judgments are based on a procedural memory
system
 Like explicit and implicit memory
 Data from remember/know experiments support
the idea that recognition is a combination of two
processes
 Recollection (remember judgments) and
 Familiarity (know judgments)
The Mirror Effect
 Observed when “The type of stimulus that is
accurately recognized as old when old is also
accurately recognized as new when new. The
type that is poorly recognized as old when old
is also poorly recognized as new when new.”
(Glanzer & Adams, 1985, p.8)
 Pervasive in recognition tests
 High/low word frequency and hit/false alarm
rates, presentation rate, age of subject, ...
The Mirror Effect - Example
The Mirror Effect and the Word Frequency Effect
Word Frequency
High
Low
Hits
27.84
31.00
False Alarms
10.20
7.63
Source: Human Memory, p. 214
The Mirror Effect
 Significance: It eliminates all theories of
recognition based on a unidimensional
conception of strength or familiarity (single
process models)
 May be able to be explained by dual process
models
 Explanations for the mirror effect are still
being formed
Face Recognition
 Face recognition versus face identification
 Other-race effect
 Face inversion effect
 Other-age effect
Face Recognition
Other-Race Effect
Face Recognition
Face Inversion Effect
Face Recognition
Face Inversion Effect
Face Recognition
Other-Age Effect
Adapted from: Human Memory, p. 220
Face Recognition
 Face recognition is closely related to expertise
with processing the stimuli
 Faces of people of the same race tend to be
recognized more accurately
 The probability of correctly identifying or recognizing
even a very familiar face decreases as it is rotated
 Young people tend to interact with young people
more and older people tend to interact with older
people more
 As with words and other stimuli, cues/priming
can be important in face recognition
Download