Lecture 4 – Psyco 350, A1 Winter, 2011 N. R. Brown Psyco 350 Lec #4 – Slide 1 Outline • Aspects of Modal Model: • Duration/Forgetting: Brown-Peterson Task • Retrieval: Sternberg Task • Problems w/ Modal Model • Dual Task Experiment • Baddeley’s Model of Working Memory – Phonological Loop – Visuo-spatial sketchpad – Central Executive Psyco 350 Lec #4 – Slide 2 Duration & Forgetting in STM • Brown-Peterson Task: Initial attempt to measure duration of STM • Procedure: – hear sub-span target set: 3 letters – count backwards for X s – recall target • Manipulation – length of retention interval • Assumption: – Counting task knocks out rehearsal – Measure of the rate of forgetting Psyco 350 Lec #4 – Slide 3 Brown-Peterson: Main Finding • In the absence of rehearsal, sub-span material is forgotten very rapidly from STM • Initial interpretation: information rapidly decays from STM • Note: w/ 0-delay, only 80% accuracy. Psyco 350 Lec #4 – Slide 4 A Test of Decay Hypothesis • Waugh & Norman (1965) -- Serial Probe Task • Method: – auditory, 16 digit list, followed by probe digit – TASK: name the digit that followed the probe • Manipulation: – location of probed item – Presentation time: 1digit/s vs 4 digits/s • Decay prediction: – recall: 1 digit/s < 4 digits/s • Interference prediction: – recall: 1 digit/s ≈4 digits/s Psyco 350 Lec #3 – Slide 5 Waugh & Norman (1965) Results: • Recall w/ # of intervening items – consistent w/ both decay & interference • Recall (more or less) unaffected by presentation rate – Consistent only w/ interference Psyco 350 Lec #4 – Slide 6 Evidence for PI in Brown-Peterson Task • Keppel & Underwood (1962) • Competing Predictions: – Decay prediction: Does delay affect recall? NO – Interference prediction: Performance decline across trials? YES • Conclusion: – Interference causes forgetting in STM Psyco 350 Lec #4 – Slide 7 Studying Retrieval from STM Psyco 350 Lec #4 – Slide 8 The Sternberg Task A Cognitive Psychology Classic Research Style: • Paradigm-driven • Exhaustive exploration of “parameter space” • Disregard for: – intrinsic importance of phenomena – individual/cultural differences – emotion & motivation Psyco 350 Lec #4 – Slide 9 Slide 9 An Example: The Sternberg Task An Information Processing Classic: • 2,500+ cites for two 1969 papers Task: Target set: short list of items Probe: a single item Target present “Old” Target absent “New”” Psyco 350 Lec #4– Slide 10 Slide 10 STM Retrieval: 3 Possibilities • Issue: Retrieval Models Parallel SelfTerminating Psyco 350 Lec #4 – Slide 11 – How do we access information in STM? – Is Item X in STM? Serial Exhaustive • Three possibilities: – Parallel – simultaneous access to all items. – Serial – consider 1 item at a time. STM Retrieval: 3 Possibilities Retrieval Models Parallel Serial Three possibilities: • Parallel – simulators access to all items. • Serial – consider 1 item at a time. – Self-terminating • Stop when: target = content SelfTerminating Exhaustive – Exhaustive • Check each item on list Psyco 350 Lec #4 – Slide 12 Selecting between Retrieval Model: The Sternberg Task • Task – Materials: • Memory Set: N letters • Probe: target letter – Question: Is probe in memory set? • Manipulations – Set Size: 1 to 6 letters – Probe Type: • positive (in memory set) • negative (not it set) Psyco 350 Lec #4 – Slide 13 Sternberg Task: Method Psyco 350 Lec #4– Slide 14 Competing Retrieval Model Predictions Psyco 350 Lec #4 – Slide 15 Why Serial Models Make Different Predictions Additive Factors Logic (Radvansky, pp. 58-60) Psyco 350 Lec #4 – Slide 16 Sternberg Task: Results • RT w/ set size Implication: serial • Negative = Positive Implication: exhaustive Psyco 350 Lec #4– Slide 17 Sternberg’s Model Psyco 350 Lec #4 – Slide 18 Problems / Serial Exhaustive Process • Conceptual: – 40 ms/comparison seems awfully fast. • Empirical: – Repetition Effect (Baddeley & Ecob, 1973): Probe = T: RTWTN < RTWGN – Serial Position Effects (Corballis et al., 1972) Probe = T: TRWGN < RTWGN Psyco 350 Lec #4– Slide 19 Alternative Approach to Sternberg Findings Assumptions: • memory set, the most active portion of LTM • memory “searched” in parallel • decision process: – “Yes”: probe-memory similarity > threshold – “No”: at deadline – similarity < threshold Set Size Effects: • encoding: activation/item as set size • retrieval: speed of assessment as set size Psyco 350 Lec #4– Slide 20 Sternberg Task: Results • RT w/ set size Implication: serial • Negative = Positive Implication: exhaustive Psyco 350 Lec #4 – Slide 21 Problems / Serial Exhaustive Process • Conceptual: – 40 ms/comparison seems awfully fast. • Empirical: – Repetition Effect (Baddeley & Ecob, 1973): Probe = T: RTWTN < RTWGN – Serial Position Effects (Corballis et al., 1972) Probe = T: TRWGN < RTWGN Psyco 350 Lec #4 – Slide 22 Alternative Approach to Sternberg Findings Assumptions: • memory set, the most active portion of LTM • memory “searched” in parallel • decision process: – “Yes”: probe-memory similarity > threshold – “No”: at deadline – similarity < threshold Set Size Effects: • encoding: activation/item as set size • retrieval: speed of assessment as set size Psyco 350 Lec #4 – Slide 23 STM, But … Model Modal accounts for serial position curve, but … not long-term recency effects. Capacity of STM is limited, but … also affected by chunking & expertise. In the absence of rehearsal, information is forgotten rapidly, but … not on the first trial. Serial exhaustive process may be used to scan STM, or… a parallel process may be used to assess the active contents of LTM. Simple “slot model” can’t account for performance of classic STM task. Psyco 350 Lec #4 – Slide 24 Dual-task Performance & the Demise of STM • STM In Modal Model – single component short-term store – functions • temporary storage – lists, task relevant info • transfer to LTM via rehearsal “A [single] system for holding and manipulating information for a wide variety of tasks such as learning, comprehension, and reasoning” -- Baddeley, p. 67 Psyco 350 Lec #4 – Slide 25 Dual-task Performance & the Demise of STM A Dual Task. 1. encode list of digits 2. perform a cognitive task (e.g., reasoning, comprehension) 3. recall list of digits. If both load and task make use of same store, then increasing load (to span) should disrupt performance on cog task. Psyco 350 Lec #4 – Slide 26 Grammatical Reason W/ Memory Load (Baddeley, 1986) • Concurrent Tasks – Memory span task – provides a memory load – Grammatical reasoning task • Procedure: – auditor presentation of digit load – 1 digit/s – concurrent overt rehearsal of load – visual presentation of letter pair & sentence – respond T/F to sentence – serial recall of digits Psyco 350 Lec #4 – Slide 27 Grammatical Reasoning Task Design: • memory load – 0 to 8 digits • sentence type – true value X voice X affirmation Psyco 350 Lec #4 – Slide 28 Implications of Single-store View • Assumptions: – Span Task – absorbs (almost) all STM capacity – Reasoning task – requires access to STM Prediction: If span task absorbs all of STM dual task requirements should produce a dramatic impairment in performance. Span-length memory load catastrophic interference Psyco 350 Lec #4 – Slide 29 Dual Task Reasoning: Results • No effect on reasoning when load is light (0-2) • reason slowed by load • error rate low regardless of load. ----------------------------------Load also has negative, but non-catastrophic, impact on free recall & text comprehension Psyco 350 Lec #4 – Slide 30 Implication of Dual Task Performance • System responsible for digit span cannot be the same as system responsible for learning / reasoning. • Motivated the development of the multicomponent WM model. Psyco 350 Lec #4 – Slide 31 Working Memory: An Alterative to STM • Baddeley and Hitch’s (1983) model • Central executive – Control center of working memory • Two slave systems: – Phonological loop • Processes verbal/acoustic information – Visuo-spatial sketch pad • Processes visual and spatial information Psyco 350 Lec #4 – Slide 32 Basic WM Model Psyco 350 Lec #4 – Slide 33 Psyco 350 Lec #4 – Slide 34 Current WM Model Psyco 350 Lec #4 – Slide 35 Dual Task Reasoning: Results • No effect on reasoning when load is light (0-2) • reason slowed by load • error rate low regardless of load. Psyco 350 Lec #4 – Slide 36 WM Interpretation of Dual Task • Load maintenance requires: – access to phonological store – minor attentional resources to schedule rehearsal • Grammatical reasoning requires: – attentional resources for sentence understanding/reasoning – limited access to phonological store – note: sentences presented visually As load , attentional demands . Thus, less capacity available for sentence processing. Psyco 350 Lec #4 – Slide 37 Phonological loop Two components • Phonological short-term store – Phonological information that decays with time • Inner ear • Subvocal rehearsal process – Articulatory-like rehearsal that needs active maintenance • Inner voice Psyco 350 Lec #4 – Slide 38 The Phonological Loop Phonological Store: holds small amount of speech based information Articulatory Control Process: Based on inner speech Phonological Store Visual Presentation Psyco 350 Lec #4 – Slide 39 Auditory Presentation Phonological Loop • Speech-based System: – phonological similarity – irrelevant speech – articulartory suppression • 2-s Capacity: – word length effect – cross-linguistic Δ’s – developmental Δ’s Psyco 350 Lec #4 – Slide 40 Evidence for the Phonological Loop Psyco 350 Lec #4 – Slide 41 Evidence for Phonological Loop • Phonological Similarity Effect similar sounding list < dissimilar sounding lists ------vs --------- • (Almost) No Evidence for Semantic Similarity Effect • Implies: representation is speech-based not meaning based. Psyco 350 Lec #4 – Slide 42 Evidence for Phonological Loop • Irrelevant Speech Effect – Recall impaired if items are accompanied by other verbal material. – Effect found w/: same-language words, samelanguage non-words, foreign words. • Interpretation: “unattended (linguistic) material was gaining access to the phonological store.” -- Baddeley, p. 53 Psyco 350 Lec #4 – Slide 43 Evidence for Phonological Loop • Articulartory Suppression – concurrent (overt or covert) articulation, decreases word span. (“the, the, the…” ; “one, two, three, one, two..) – concurrent articulation decreases • the phonological similarity effect • word length effect. • Interpretation: articulation of irrelevant items dominates ACP - Words cannot be “rehearsed” or recoded into phonological code Psyco 350 Lec #4 – Slide 44 Evidence for the Phonological Loop Psyco 350 Lec #4 – Slide 45 Evidence for Phonological Loop • Word Length Effect word span decrease as # of syllables/word increases. • Recall depends of reading rate. – # words recalled ≈ 2 * (reading rate) • reading rate = # words read / s Psyco 350 Lec #4 – Slide 46 Phonological Loop – Capacity of Phono Store Baddeley et al (1975) Task: serial recall Materials:5-word lists Manipulation: syllable length Results: • recall , as syllable length • recall predicted by reading rate. • cf. STM predictions Psyco 350 Lec #4 – Slide 47 Phonological Loop – Capacity of Phono Store Baddeley et at (1975) • linear relation between reading time & recall • Interpretation: – capacity of phono loop ~ 2 s of speech materials • Reason – fast fading phono trace – rehearsal refreshes trace. – if not rehearsed within 2 s, most info lost. Psyco 350 Lec #4 – Slide 48 Phonological Loop – Capacity of Phono Store Baddeley et at (1975) • Implications: – across languages, digit span should be related to mean syllable length of digits. – digit span should increase w/ age, because speech rate does. Psyco 350 Lec #4 – Slide 49 Cross-Linguistic Δs in Digit Span Naveh-Benjamin & Ayers (1986) As predicted: • span larger for languages w/ short digits than long • span ≈ 2 X reading rate Psyco 350 Lec #4 – Slide 50 Age-related Δs Digit Span Hulme (1984) As predicted: • span w/ age • span ≈ 2 X speech rate ----------------------Overt or covert articulation serves to maintain items in the phonological store by refreshing their fading traces. The faster it can run, the longer the memory span Psyco 350 Lec #4 – Slide 51 Phonological Loops: Functions • Learning to read: Children with impaired reading ability have reduced memory spans and have difficulties in tasks which require the manipulation of phonological information (e.g. given Stop, reply Top). • Language comprehension: STM patients some difficulty in comprehending verbose or complex sentences e.g. “The boys pick the apples” = OK; “The two boys pick the green apples from the tree” = Impaired • Vocabulary acquisition There is a strong correlation between non-word repetition (which strongly taxes the phonological loop) and vocabulary size (Gathercole & Baddeley, 1989) Psyco 350 Lec #4 – Slide 52 Psyco 350 Lec #4 – Slide 53 VSSP • Function: – “construction, maintained, & manipulation of mental images.” – Radvansky, p. 97 • Assumptions: – Independence of VSSP & Phonological Loop Psyco 350 Lec #4 – Slide 54 Visuo-spatial sketchpad Operations: • Mental rotation • Mental scanning • Boundary extension • Dynamic memory Supports: Spatial problem-solving (moving a couch) Prediction of dynamic consequences. Psyco 350 Lec #4 – Slide 55 Independence of VSSP & Phonological Loop Brooks (1968) • Dual Task Experiment – Goal to demonstrate: • spatial response mode interference w/ spatial processing • verbal response mode interferes w/ verbal processing • cross-modal tasks produce little interference Psyco 350 Lec #4 – Slide 56 Independence of VSSP & Phonological Loop Brooks (1968) • Design (2X3) TASK X image scanning grammatical decision Psyco 350 Lec #4 – Slide 57 RESPONSE MODE pointing tapping vocal Brooks (1968): Image Scanning Task Given a block letter & starting point: If current corner is top or bottom “yes” Otherwise ”no” Psyco 350 Lec #4 – Slide 58 Brooks (1968): Image Scanning Task Given a proverb/cliché: If current word is noun “yes” Otherwise ”no” Example: Psyco 350 Lec #4 – Slide 59 Brooks (1968): Response Modes While performing target task: • Vocal – say “yes”/”no” • Taping – left tap = “yes”; right tap = “no” • Pointing – point to successive “y”/”n” pairs on response sheet Psyco 350 Lec #4 – Slide 60 Brooks (1968): Pointing Psyco 350 Lec #4 – Slide 61 Brooks (1968): Results >> ≈ • Spatial response mode: – interfered w/ spatial task – did not interfere w/ verbal task • Verbal response mode: – interfered w/ verbal task – did not interfere w/ spatial task Psyco 350 Lec #4 – Slide 62 ≈ << Brooks (1968): Interpretation • Task X Mode interaction indicates: – separate & limited pool of resources for verbal & spatial task • image scanning task & spatial response mode draw on the limited resources of the VSSP. • grammatical decision task & vocal response mode both draw on the resources of the phonological loop. Psyco 350 Lec #4 – Slide 63