PSYC 303 2024F Learning - PDF

PSYC 303 | L05 LEARNING Oliver Hardt PhD McGill University/Department of Psychology The University of Edinburgh/Medical School/The Patrick Wild Centre (PWC) [email protected] https://www.mcgill.ca/psychology/oliver-hardt http://patrickwildcentre.com/oliver-hardt OVERVIEW 1. Introduction 2. Fac...

PSYC 303 | L05 LEARNING Oliver Hardt PhD McGill University/Department of Psychology The University of Edinburgh/Medical School/The Patrick Wild Centre (PWC) [email protected] https://www.mcgill.ca/psychology/oliver-hardt http://patrickwildcentre.com/oliver-hardt OVERVIEW 1. Introduction 2. Factors determining learning success 3. Types of learning 4. Neurobiology memory formation INTRODUCTION 1  4 NATURE VS NURTURE WHAT IS THE ROLE OF LEARNING IN OUR LIVES? ‣ Nativism. Knowledge of the world is mostly innate INTERACTIONISM VIEW (a priori), and determines certain abilities (Descartes was a nativist). This view is associated with the idea that NATURE (i.e., genes etc) determines behaviour. ‣ Empiricism. At birth mind is tabula rasa, nothing in terms of behaviour and knowledge is inherited, all 1 INTRODUCTION is learned (Locke was an empiricist; he also never had children). This view is associated with the idea that NURTURE (environment) determines behaviour. ‣ Interactionism. Certain basic abilities and knowledge is innate, but can be influenced by experience, which in turn can change what innate behaviours are expressed, which influences what will be experienced, and so on. This view is supported Claridge & Davis (2003) strongly by the available evidence.  5 NATURE VS NURTURE THE ROLE OF TWIN STUDIES TO DETERMINE INFLUENCE OF NATURE AND NURTURE ‣ Genotype. The complete set of an organism’s genes. ‣ Phenotype. The overt characteristics of an organism, i.e., the properties we can observe (resulting from what genes are expressed). Phenotype is product of genotype and 1 INTRODUCTION environmental factors, and experience. ‣ Example: human skin colour. Our genes determine how much melanin we produce, but exposure to UV-B radiation can increase melanin production, which darkens the skin because melanin is black. ‣ Example: Flamingos are white but the pigments in the animals they eat causes their feathers to turn pink.  6 NATURE VS NURTURE THE LOGIC OF TWIN STUDIES TO UNDERSTAND NATURE VS NURTURE IN HUMANS Dizygotic (fraternal) twins Monozygotic (identical) twins 1 INTRODUCTION  7 NATURE VS NURTURE THE LOGIC OF TWIN STUDIES TO UNDERSTAND NATURE VS NURTURE IN HUMANS ‣ The study design recruits twins adopted into different families. ‣ They share their genes from their biological parents, but grow up in different environments with the adoptive parents, with whom they do not share 1 INTRODUCTION genes. ‣ Finding different combinations of monozygotic and dizygotic twins growing up adopted and with their own parents, as well as control children allows to determine the influence of nature (genes) and nurture (family environment) on cognitive function (e.g., intelligence, memory, etc.).  8 NATURE VS NURTURE POSSIBLE GROUPS AND WHAT THEY CAN TELL US ABOUT TRAITS Study design: groups required to examine to what extent performance on IQ tests reflects the influence of inherited traits or environmental factors during childhood development. ‣ Monozygotic twins growing up together: Provides data for same “nature”, similar “nurture” ‣ Monozygotic twins growing up apart: Provides data for same “nature”, different “nurture” ‣ 1 INTRODUCTION Dizygotic twins growing up together: Provides data for similar “nature”, similar “nurture” ‣ Dizygotic twins living apart: Provides data for similar “nature”, different “nurture” ‣ Siblings growing up together: Provides data for less similar “nature”, similar “nurture” ‣ Individuals not related to each other living apart: Baseline control group. Provides data for genetically unrelated, separately living individuals that should correlate in IQ scores randomly. WHAT DETERMINES SUCCESS IN LEARNING? 2  10 TOTAL TIME SPENT LEARNING THE TOTAL TIME HYPOTHESIS 2 INFLUENCES ON LEARNING ‣ Ebbinghaus learned lists containing 16 nonsense syllables. He varied how often he recited the list (from 8-64 repetitions). Then he measured time required to relearn the list the next day. ‣ He found monotonous linear relationship between the number of recitations and time required to relearn the list: The more recitations on day 1, the faster relearning on day 2. ‣ Total time hypothesis: the more time you spend learning, the higher the amount learned. In other words: A set amount of time is required to learn a specific amount of material, no matter how many study sessions that time is spread across.  11 TOTAL TIME SPENT LEARNING EXPERTISE AND DELIBERATE PRACTICE 2 INFLUENCES ON LEARNING ‣ According to the total time hypothesis, you could gain expertise in many areas provided you would practice/learn long enough. ‣ This lead to the idea of the “10k hour” rule: studies suggest that mastery in any field (e.g., outstanding musicians, architects, dancers, memory scientists …) requires around 10k hours of practice. However, there is large variety in the data — some masters spend 5k hours, others 25k hours. ‣ Total time of learning/practice alone cannot explain mastery. There is something else. ‣ One source is “nature” (see introduction); ‣ the other how learning/practice is done. ‣ Deliberate practice. A targeted learning activity consisting of ‣ Identify particular weakness in a skill, and focus on the weakness. ‣ Training intends to improve this, and immediate feedback is provided. ‣ Gradual refinement favoured over simple repetitions, which is not enough once you reached a certain level of expertise.  12 DISTRIBUTED PRACTICE THE REPETITION EFFECT 2 INFLUENCES ON LEARNING ‣ Participants saw a list of words. Some of these words occurred once, others twice. ‣ It varied how long words were presented (from 1.3 sec to 4.3 sec). ‣ It also varied how many other words were between the words that were shown twice (from 0 to 40). ‣ Participants then recalled as many words as possible in a free-recall test. ‣ Results show ‣ Increasing presentation rate improves free recall. ‣ Repetition effect: Repeated words better recalled than non-repeated. ‣ Lag effect: Increasing the distance between words reoccurring improves their free recall. Melton 1970 J Verb Learn Verb Behav 9  13 DISTRIBUTED PRACTICE HOW RELEVANT ARE THESE RESULTS FOR REAL WORD SCENARIOS? 2 INFLUENCES ON LEARNING ‣ In an online study, participants first learned 32 obscure trivia facts until they could answer each of the trivia questions. ‣ They relearned these facts at after lags of various lengths (0-105 days). ‣ Then they took a memory test on the trivia questions after different retention intervals (7-350 days). ‣ Results show a non-monotonic lag effect: ‣ Worst is no lag (relearning it immediately; i.e., massed practice). Best effect on memory performance with lags around 10-20 days. ‣ The longer the retention interval, the worst memory performance. ‣ Lag effect smallest for shortest (7 days) and longest (350 days) retention interval. Cepeda et al. 2008 Psychon Bull Rev 132  14 DISTRIBUTED PRACTICE THEORETICAL EXPLANATIONS OF THE LAG EFFECT 2 INFLUENCES ON LEARNING ‣ Deficient processing hypothesis. You tend to pay less attention to recently encountered stimuli (they are easier recognized as familiar). Therefore these items will be less processed and not benefit much from repetition. ‣ Encoding variability hypothesis. The more different you encode a stimulus, the better it will be retained. This relates somewhat to the dual-trace hypothesis on distributed practice discussed in L01 (How to improve memory). The longer you wait between repetitions, the more likely learning will lead to encoding of stimulus in different context, which provides additional retrieval cues etc. ‣ Study-phase retrieval hypothesis. In a test for an item, you will retrieve it (the “study phase”), and retrieval itself will produce a strengthening of the item representation. The more effort (the more difficult) retrieval, the stronger the effect on it. Thus, longer lags between retrievals likely will result in more difficult retrieval, which can have a stronger effect on memory retention than shorter intervals.  15 DISTRIBUTED PRACTICE THEORETICAL EXPLANATIONS OF THE LAG EFFECT 2 INFLUENCES ON LEARNING TWO CONTRASTING THEORETICAL PREDICTIONS ‣ If the encoding variability hypothesis is correct, then better remembered repeatedly presented items should go along with higher variation in brain activity patterns in areas supporting memory for a stimulus. ‣ If the study-phase retrieval hypothesis is correct, better memory should go along with higher pattern similarity (less variance) in these brain regions. ‣ In a study to test this, 24 participants saw in a scanner 120 novel faces. Each face was presented four times, with lag ranging from 1 (i.e., consecutive) to 20 faces. ‣ One hour after the scan, participants took a yes/no recognition memory test (using a response scale from 1=definitely new to 6=definitely old), during which a total of 240 faces (half old, half new) were randomly mixed together. ‣ Behavioural Results. Out of the 120 old faces, subjects on average recognized with high confidence (e.g., a 5 or 6 rating) 51.7 ± 18.6 items and forgot (a 1 or 2 rating) 37.3 ± 16.9 items. Xue et al. 2010 Science 330  16 DISTRIBUTED PRACTICE THEORETICAL EXPLANATIONS OF THE LAG EFFECT 2 INFLUENCES ON LEARNING ‣ Using multi-voxel pattern analysis, they tried to detect similar activity patterns in regions of interest (ROIs) in the brain (regions associated with visual memory). ‣ They did this separately for Remembered faces (R) and Forgotten faces (F). ‣ The study-phase retrieval hypothesis would predict that when faces are remembered (R), the patterns are more similar than when they are forgotten (F). ‣ The encoding variability hypothesis would predict the opposite. Xue et al. 2010 Science 330  www.science.org at Mcgill University Library on September 19, 2024 the scan (10). To prevent further encoding of each Fig. 2. Neural pattern similarity in a sample region. (A) The location of the right dorsal lateral occipital item during the repetition lag, subjects performed 17 DISTRIBUTED PRACTICE cortex (RdLOC), which was anatomically defined according to the Harvard-Oxford probabilistic map, and overlaid onto the group-averaged anatomical map. (B) Neural pattern similarity from a single subject’s single-run data. Pattern similarity was calculated by computing the correlation between the parametric a highly engaging self-paced visual orientation judgment task for 8 s after each semantic judgment estimates (beta) for each voxel within the ROI across the two repetitions. The line reflects unit slope. (C) task (lasting for 3 s), and the next trial started after THEORETICAL Neural EXPLANATIONS pattern similarity averaged OF across all subjects (n = 24), THEforLAG separately each pairEFFECT of a repetition a 1-s delay (fig. S6). Each item was repeated three combination. (D) The mean neural pattern similarity as a function of subsequent memory. A repeated- times, with a repetition lag ranging from four to 2 INFLUENCES ON LEARNING measures analysis of variance (ANOVA) was used to examine the differences between conditions. Error nine trials. Thirty minutes after the study session, ‣ bars represent participants were asked to freely recall the words rightwithin-subject inferior error. REM,rightremembered; ventral FORG, forgotten. lateral bilateral ventral They they had studied in thefound scanner. Outthat of thesimilarity 60 items, of parietal lobule occipital cortex visual cortex activity subjects, on average, pattern recalled 13.5 Tin 4.7regions items (table S1). where face stimuli are Subjects’ response time on the semantic judg- processed ment task decreased predicts(F2,42 across repetitions better = memory 42.96, P < 0.001); accuracyforwasthem. high (mean = 97.5% T 2%) and did not change across repeti- ‣ tions (F2,42 =Forgetting faces is(Fassociated 1.68, P = 0.19). Accuracy 1,21 = 1.79, P = 0.19) and response with times (F1,21 less pattern = 0.15, similarity. P = 0.70) did not differ between subsequently ‣ recalled itemsThisand outcome forgotten items is (table not predicted S2). There were no bysignificant the encodinginteractions variability between repetition and subsequent memory for either ac- hypothesis, but supports curacy (F2,42 = 0.14, P = 0.087) or response time the (F2,42 = 1.69,study-phase P = 0.20). Functional retrieval imaging data revealed that,hypothesis. similarly to Experiment 2, there were significantly stronger activations for sub- sequently recalled items than for subsequently forgotten items in the left middle and inferior frontal gyrus (LMFG/LIFG) (MNI: –50, 14, 34; Pattern similarity Z = 4.17), and the left dorsal lateral occipital Fig. 3. Neural patterninsimilarity regionsis of interest associated (ROI) with for recalled face memory. Greater(R) and similarity pattern forgotten for (F) faces.remembered faces than for subsequently forgotten faces was found in (A) the right inferior subsequently lobe (LdLOC) and adjacent inferior parietal lobule parietal lobule (RIPL), (B) the right ventral lateral occipital cortex (RvLOC), which were anatomically (LIPL) (MNI: –40, –66, 46; Z = 3.48) (fig. S7). defined, and2010 (C) theScience bilateral 330 ventral visual cortex, which includes the bilateral fusiform gyrus, bilateral We then examined whether the degree of pat- Xue et al. tern similarity in the 20 anatomically defined re- inferior temporal gyrus and bilateral ventral lateral occipital cortex, but excludes voxels showing signif- gions was associated with subsequent memory  18 DISTRIBUTED PRACTICE CONCLUSION ON DISTRIBUTED PRACTICE 2 INFLUENCES ON LEARNING ‣ Repeated study improves memory retention. ‣ But: distributed practice is always better than massed practice. ‣ The effect between extent of distribution (lag) and positive effect on retention is not linear: at some point, there is no more increase in retention with increased lag. ‣ Distributed practice can effectively increase coherence of brain activity patterns in areas important for memory of the studied material.  ptember 19, 2024 ptions regard- g and repeated 19 once informa- , what are the RETRIEVAL-BASED LEARNING Table 1. Conditions used in the experiment, average number of trials within each study or test period, and total number of trials in the learning phase in each condition. SN indicates that only vocabulary pairs not recalled in the previous test period were studied in the current study period. TN g study trials) ials) on learn- indicates that only pairs not recalled in the previous test period were tested in the current test essed after a THE TESTING period. Students EFFECT in all conditions performed a 30-s distracter task that involved verifying multi- plication problems after each study period. 2 INFLUENCES ON LEARNING this research ments of their t of materials Condition Study (S) or test (T) period and number of trials per period Total number ‣ Participants studied list of 40 est conditions, 1 2 3 4 5 6 7 8 of trials Swahili-English word pairs (e.g., r future recall mashua-boat). ST S T S T S T S T Our question ight into their 40 40 40 40 40 40 40 40 320 ‣ They were tested with the cue and SNT S T SN T SN T SN T had to produce the target (e.g., iment was to learning and 40 40 26.8 40 8.0 40 2.0 40 236.8 mashua-?). urdue University, STN S T S TN S TN S TN ‣ Once a word pair was recalled ment of Psychol- 40 40 40 27.9 40 11.8 40 3.3 243.0 correctly, it was treated , St. Louis, MO SNTN S T SN TN SN TN SN TN differently in the four conditions. dressed. E-mail: 40 40 27.1 27.1 8.8 8.8 1.5 1.5 154.8 ‣ ST: Participants studied always all words and were always tested on all words. ‣ SNT: Once a pair was recalled, it was dropped from further study but tested in each subsequent test 15 FEBRUARY 2008 VOL 319 SCIENCE www.sciencemag.org period (SNT indicates that only non-recalled pairs were restudied). ‣ STN: Recalled pairs were dropped from further testing but studied in each subsequent study period. ‣ SNTN: Recalled pairs were dropped from study and test periods. Many students do this – study something until it is learned (i.e., can be recalled) and then drop it from further practice. Karpicke & Roediger 2008 Science 319  20 RETRIEVAL-BASED LEARNING THE TESTING EFFECT 2 INFLUENCES ON LEARNING ‣ The traditionally “recommended” method (SNTN) had worst benefit for memory of them all. ‣ ST and SNT had best effects. ‣ SNT, however, is more efficient because fewer total number of trials is required. ‣ What does that mean for you? If you wish to study for an exam, study those items you cannot recall, but continue to test yourself on all items that you have to remember, even if you can recall them perfectly. The best way to do this would be a free recall procedure (difficult, i.e., more retrieval effort). ‣ However: for longer retention (time periods > 7 d), studying items again will help, but as deliberate practice. Karpicke & Roediger 2008 Science 319  21 RETRIEVAL-BASED LEARNING ROLE OF RETRIEVAL EFFORT ON TESTING EFFECT 2 INFLUENCES ON LEARNING ‣ The retrieval effort hypothesis states that difficult but successful retrievals are better for memory retention than easier successful retrievals. ‣ This study tests the hypothesis: Participants learned 10 lists each containing 7 Swahili-English word- pairs. After they learned all word-pairs, a cued recall test was administered. ‣ Independent variables: ‣ Criterion (within subject): each item on the list was assigned to a retrieval criterion (1-10), indicating how often an item has to be correctly retrieved during training before it will be dropped from training. Participants knew training went on until an “acceptable” performance was reached, but did not know about the actual criteria. ‣ Inter-stimulus interval (ISI; number of word-pairs between two repetitions of same word pair; between subjects): ‣ short ISI: participants studied a list of 7 pairs until criteria were reached, then the next list. ISI was 6 items, around 1 min. ‣ long ISI: participants studied the first 5 lists together (35 pairs), then the last 5 lists. ISI was 34 items, around 6 min. ‣ Retention interval (RI, between subjects): time interval between end of study and test, 25 min (short) vs 7 d (long). Pyc & Rawson 2009 J Mem Lang 60  22 RETRIEVAL-BASED LEARNING ROLE OF RETRIEVAL EFFORT ON TESTING EFFECT 2 INFLUENCES ON LEARNING ‣ Retrieval effort manipulation Long ISI ‣ long ISI > short ISI 25 min RI the more time passes between two retrieval attempts, the more retrieval Short ISI effort is required. ‣ Criterion 10 > Criterion 1 Long ISI the more often an item is correctly retrieved, the easier future retrievals, but the effect lessens with each successful 7 d RI retrieval (effort gets lesser and lesser). Short ISI ‣ Results Confirm Retrieval Effort Hypothesis ‣ Memory for long ISI > short ISI ‣ Memory for Criterion 10 better than 1 ‣ Criterion effect weakens progressively ‣ Memory better for short than longer RI Pyc & Rawson 2009 J Mem Lang 60  23 RETRIEVAL-BASED LEARNING TEST-ENHANCED LEARNING 2 INFLUENCES ON LEARNING ‣ Test-enhanced learning is the fact that studying an item after it has been retrieved (i.e., memory for it has been tested) enhances its memory retention. ‣ Participants read a text passage and either received no test, or multiple- choice tests with or without immediate or delayed (at end of test) feedback. ‣ Participants were tested 7 d later. ‣ Results show test-enhanced learning: ‣ Testing without feedback tripled recall performance. ‣ Testing with delayed feedback most effective (5 times better than without testing). Butler & Roediger 2008 Mem Cog 36  24 RETRIEVAL-BASED LEARNING CONCLUSION ON RETRIEVAL-BASED LEARNING 2 INFLUENCES ON LEARNING ‣ Memory retention benefits more from testing all material than from studying all material. ‣ The more challenging the test, i.e., the more difficult the retrieval approach (e.g., free recall), the stronger the effect on memory retention. ‣ Feedback on retrieval outcome (what is correct and what is not) dramatically increases memory retention. ‣ Receiving delayed feedback can lead to stronger effects than immediate feedback.  25 MOTIVATION TO LEARN TWO BASIC FORMS OF MOTIVATION 2 INFLUENCES ON LEARNING ‣ Intrinsic motivation ‣ Motivation that is not based on reward/punishment, but on outcome-independent curiosity, sense of accomplishment, need for competence. ‣ Extrinsic motivation ‣ Motivation that is based on external rewards or punishments, for example to perform behaviour to gain money, grades, praise, recognition, or to avoid disapproval. ‣ Is learning outcome (i.e., memory retention) affected by intrinsic or extrinsic motivation?  26 MOTIVATION TO LEARN LEARNING AND INTRINSIC MOTIVATION 2 INFLUENCES ON LEARNING ‣ Participants rated how curious they were about the answers (assumption: more curious = higher intrinsic motivation to learn answer) ‣ Results ‣ Memory for answers was better for high-curiosity questions (15-20% better). ‣ Faces presented after high- curiosity questions were remembered better. ‣ High-curiosity questions went along with higher activity in VTA and nucleus accumbens. ‣ VTA is one major source of dopamine in brain. Gruber et al. 2014 Neuron 84  27 MOTIVATION TO LEARN LEARNING AND INTRINSIC MOTIVATION 2 INFLUENCES ON LEARNING ‣ Both VTA and nucleus accumbens (NAc) are involved in reward processing and motivational aspects of memory formation. ‣ VTA is the second major source of dopamine in the human brain. VTA sends dopaminergic projections to hippocampus, and NAc. ‣ NAc provides excitatory (glutamatergic) input to hippocampus. ‣ In response to curiosity, input from both areas enhance synaptic plasticity in the hippocampus. ‣ This improves memory formation for events surrounding their signalling, such that trivia questions and faces following them are better remembered. Image Credit: George V Kach - - , 28 MOTIVATION TO LEARN - t LEARNING AND EXTRINSIC MOTIVATION 2 INFLUENCES ON LEARNING d t ‣ Reward-based learning. Offering rewards (praise, better grades, money, g etc.) for item retention improves long- - term memory for them. l , ‣ The neural basis for this effect is e explored in the study shown on the right. - ‣ In a scanner (fMRI), participants were - promised a high or low monetary reward - for remembering photos of scenes. s Figure 1. Task Trial Structure ‣ A cue announced the reward first (low or d High-value trials are depicted for the monetary incentive delay (MID) high), then the to-be-remembered item - task (top) and the monetary incentive encoding (MIE) task (bottom). was presented. Recognition memory was Gradient bars represent BOLD signals modeled in analyses of each tested about 24 hours later. e interval. In both tasks, a cue indicated the value of each upcoming n target. In the MID task, the correct response was a button press ‣ Behavioural Result. Recognition accuracy l during the rapidly presented target (i.e., a white square). This was was higher for high-reward than low- - followed by feedback about reward and cumulative earnings. In reward scenes. - the MIE task, the correct response was recognition of the target sceneAdcock et al. at stimulus 206test Neuron w2450hr later. Scenes in the MIE task were.  29 MOTIVATION TO LEARN CONCLUSION ON ROLE OF MOTIVATION IN LEARNING 2 INFLUENCES ON LEARNING ‣ Both intrinsic and extrinsic motivation affect learning. ‣ Higher motivation affects dopamine signalling in the brain, which promotes long-term memory formation in areas such as the hippocampus. ‣ Both intrinsic and extrinsic motivation engage overlapping neuromodulatory systems (e.g., VTA and NAc).  30 AMOUNT OF ATTENTION DURING LEARNING DIVIDING ATTENTION IMPAIRS LONG-TERM MEMORY RETENTION 2 INFLUENCES ON LEARNING ‣ Dividing attention impairs task performance. Whether dividing attention also affects memory retention was tested in the study on the left. ‣ Electrodes were attached to participants who were shown 12 words (as a ruse). Both incidental learning and intentional learning group were told that the electrodes measured reactions to the word. Only the intentional learning group was asked to learn words for later test. ‣ While words were shown, participants in the divided attention condition (DA) judged whether a tone played at the same time was low, medium, or high pitched (like smartphone notification). ‣ Then participants took a free recall test. ‣ Results: DA significantly impaired memory in both incidental and intentional learning conditions. Naveh-Benjamin & Brubaker 2019 J Mem Lang 106  31 AMOUNT OF ATTENTION DURING LEARNING POSSIBLE EXPLANATIONS FOR EFFECTS OF DIVIDED ATTENTION ON MEMORY 2 INFLUENCES ON LEARNING ‣ The (left) ventrolateral prefrontal cortex has been implicated to affect activity of the hippocampus: It can enhance hippocampal activity, which can facilitate memory encoding processes. ‣ This can improve persistence of memory. ‣ Dividing attention during learning tasks reduces activity of (left) ventrolateral prefrontal cortex. ‣ Therefore, divided attention can interfere with activity of the ventrolateral prefrontal cortex, which in turn can impair encoding of new memories in the hippocampus, reducing memory persistence.  32 THE ROLE OF SLEEP IN LEARNING THE MEMORY CONSOLIDATION CONCEPT 2 INFLUENCES ON LEARNING ‣ Müller & Pilzecker completed a series of experiments based on the methods Ebbinghaus introduced, using new apparatuses like the one on the left. ‣ Participants learned lists of nonsense syllables and memory for them was tested in free recall. ‣ They observed that participants tended to reiterate the syllables they were asked to remember (“perseveration”), and that disruptions of this process impaired memory for the syllables. ‣ This lead to the perseveration-consolidation theory that became one of the fundamental concepts in memory research studied to this day.  33 THE ROLE OF SLEEP IN LEARNING THE MEMORY CONSOLIDATION CONCEPT 2 INFLUENCES ON LEARNING ‣ Perseveration-Consolidation hypothesis “The tendency to perseverate [...] might serve to consolidate the associations among [the syllables].” Müller & Pilzecker 1900 p. 68 ‣ Current consolidation “dogma” McGaugh 1966 Science 153; 2000 Science 287 ‣ Memories are labile after acquisition and are fixed, i.e., permanently stored (consolidated) over time. ‣ Consolidated memories are stable and can persist long-term (cf. Ribot’s Law of Regression). ‣ Consolidation is a transient, unidirectional process that occurs after acquisition. ‣ Disrupting consolidation impairs memory retention.  the case of Me, the averagesat the 4 and 8 hr. intervals increase slightly; and similarly, for H at the 8 hr. interval, the average is slightly greater than at the 2 hr. interval. 34 THE ROLE OF SLEEP IN LEARNING Comparativecurves are shown in Fig. I, in which the average number of syllables reproducedafter the various time-intervals is plotted separ- ately for each 0 and for the sleep and waking experiments. The curves of the waking experiments take the familiar form: a sharp decline which becomes progressively flatter. The form of the curves of the sleep ex- periments, however, is very different: after a small initial decline the curves flatten and a high and constant level is thenceforth maintained. THE ROLE OF SLEEP IN MEMORY: AN EARLY TEST OF THE CONSOLIDATION THEORY These results are not due to a differencein the depth of learning; for, as far as we were able to determine, this was constant for every experi- 2 INFLUENCES ON LEARNING ment: learning was just brought to complete mastery; repetitions ceased at the first correct recitation. The statement that the depth of learning ‣ Participants (N=2) learned lists of nonsense syllables. ‣ After learning participants either went to sleep (learning was in the evening) or went about their day at college. ‣ Memory of participants were tested in free recall for varying retention intervals (0, 1, 2, 4, 8 hours). ‣ Pr4 CQ