Methods In Memory Research Lecture Notes PDF

PSYC 303 | L03 METHODS IN MEMORY RESEARCH 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. I...

PSYC 303 | L03 METHODS IN MEMORY RESEARCH 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. Experiments 3. Measuring Memory 4. Measuring brain function INTRODUCTION 1  4 TWO BASIC POSITIONS DIFFERENCE BETWEEN RATIONALISM AND EMPIRICISM ‣ “Psychology has a long past, but a short history” (Hermann Ebbinghhaus) ‣ Psychological inquiry may be as old as humanity, at least as old as recorded history. Questions about memory have been advanced over millennia. In this sense, psychology has a long past. ‣ Rationalism: For the most part, psychological questions were framed and pursued within the framework of rationalism – using reason & logical 1 INTRODUCTION arguments was thought best to find the correct answers about the nature of mind. Rationalism argues that observation not only is unnecessary, but potentially misleading. ‣ For example, using reason, Aristotle argued that the brain was part of the cooling system of the body, and thus has little to do with memory. ‣ Empiricism: Hypotheses and theories about memory need to be confirmed or disconfirmed by observation. There are several empirical methods permitting proper observation, descriptive methods as well as experimental methods. ‣ As an empirical science, psychology (and thus memory) has a short history.  5 THE EMPIRICAL APPROACH COMPONENTS AND GENERAL PROCEDURE OF A PROPER SCIENCE Empirical Methods 1 INTRODUCTION  6 THE EMPIRICAL APPROACH COMPONENTS AND GENERAL PROCEDURE OF A PROPER SCIENCE ‣ Observations. Ideally unbiased, but it is questionable if there are unprejudiced observations. We usually have an idea for what we are looking for, and observe using judgements and concepts (prior knowledge). Our observations are therefore not purely objective, but also subjective. ‣ Data. Data do not equal facts. They usually are produced 1 INTRODUCTION by measurements and quantifications. These need to be interpreted within a theory: Facts do not exist objectively, but emerge from a context. Fact=Data+Theory (Deese, 1972). ‣ Theory. Scientific theories and research reflect biases, prejudices, values, assumptions, history of the individual scientists and of the scientific community, which is embedded in a cultural context. Science is not free of values.  7 LOGIC OF TESTING PREDICTIONS TWO PRINCIPAL APPROACHES TO TESTING THEORIES ‣ Classic Method: Inductivism. Several observations are used to induce theories, which are used to deduce hypothesis, leading to more observations, eventually to arrive at laws. Example: observing sufficient white swans might lead to law that “ALL SWANS ARE WHITE”. ‣ Falsificationism (Popper’s suggestion). A statement (hypothesis) that is capable of being refuted is deduced from 1 INTRODUCTION a theory. Tests are designed to refute the predictions (falsification), not to confirm a theory. ‣ Theories allow for an infinite number of predictions. It is impossible to confirm all of them. ‣ But if one prediction is falsified, then theory is wrong. ‣ Therefore, falsificationism allows for a strict testing of theories. ‣ If a prediction of a theory is falsified, theory can be changed and tested again. EXPERIMENTS 2  9 CAUSE AND EFFECT RELATIONSHIP THE EXPERIMENTAL METHOD ‣ Scientists are interested in explanation for phenomena, in establishing cause and effect relationships. ‣ The experimental method is a way to detect such causalities. ‣ In an experiment, in order to test a hypothesis, experimenters manipulate one or several variables, and, while trying to keep everything else constant (i.e., the same; ceteris paribus 2 THE EXPERIMENT principle), they observe the effect of this manipulation on one or several other variables (sometimes called outcome variables). ‣ Definition: An experiment is a research method that tests causal hypothesis by manipulating and measuring variables. ‣ The manipulated variables are called independent variables (IV), the variables to measure the effect of the manipulations are called dependent variables (DV).  10 BASIC COMPOMENTS OF EXPERIMENTS TWO PRINCIPAL GROUPS IN AN EXPERIMENT ‣ Experimental group(s). Participants who receive or have received a treatment. For example, when you want to compare effects of brain damage on memory retention, the participants with brain damage are technically in a “treatment” group. ‣ Control group(s). Participants who receive no treatment or 2 THE EXPERIMENT who receive a treatment that is unrelated to the independent variable being investigated. For example, if you are interested in how testing affects memory retention, the ones receiving no testing are the control group. ‣ This design of a study allows to detect causal relationships between the IV (e.g., brain damage; being tested) and the DV (memory retention). If the IV consistently influences the DV, then the IV is assumed to cause changes in the DV, assuming all other things being equal (ceteris paribus).  11 FLOW CHART OF EXPERIMENTS OVERVIEW OF THE EXPERIMENTAL METHOD 2 THE EXPERIMENT  12 TWO BASIC DESIGN PLANS BETWEEN SUBJECTS VS REPEATED MEASURES (WITHIN SUBJECTS) DESIGN BETWEEN SUBJECTS DESIGN WITHIN SUBJECTS DESIGN 2 THE EXPERIMENT ‣ Advantages of between subjects ‣ Advantages of within subjects designs: designs: each participant only subjects (a) fewer participants are needed than to one treatment condition. The DV in a corresponding between-subjects therefore less influenced by (a) practice/ design; (b) reduces measurement errors experience effects; (b) fatigue/boredom because individual differences between of participants; (c) sequence effects. groups of subjects are eliminated. This design is versatile and can be used for many research questions. ‣ Disadvantages: (a) progressive error, because performance is also influenced ‣ Disadvantages: (a) relatively large by experience of repeatedly number of participants needed; (b) participating (e.g., fatigue); (b) measurements are obtained from carryover effects, treatment in a different participants, inducing variance particular condition might change due to their different characteristics. participants such that it affects performance in a later condition (e.g., practice effects).  13 TECHNIQUES FOR GOOD EXPERIMENTS FACTORS AFFECTING INTERNAL VALIDITY (QUALITY) OF EXPERIMENTS ‣ In order to allow causal inferences, experiments depend on rigorous control: Ceteris paribus principle (i.e., everything else the same except IV). ‣ Control. All steps taken by the experimenter to minimize the possibility that other variables besides the IVs influence the DV. ‣ Confound. Anything that affects a DV and that varies unintentionally 2 THE EXPERIMENT between the IV levels. Examples: time of day, sequence of items presented, educational level of participants; etc. ‣ Random assignment. Participants in the sample are randomly assigned to the experimental conditions (i.e., the levels of the IV). Each participant in the sample has the same chance being assigned to any treatment level. Random assignment ensures that on average individual differences of participants are comparable between experimental groups. ‣ Double Blind Procedure. Neither the experimenter running the experiment nor the participant should know about the treatment condition of the participant. This eliminates that knowledge of the treatment group influences performance of the participant (a well-documented effect in human and animals studies). MEASURING MEMORY 3  15 HOW TO MEASURE MEMORY? MEMORY CANNOT BE DIRECTLY OBSERVED ‣ Memory is never directly observed. In humans, awareness of its influence on behaviour not always present (e.g., priming effects). ‣ No agreement as to what counts as memory. From a biological perspective, every change in neural connectivity reflects memory. Chronic pain, for example, is memory in the sense as it depends on long-lasting changes in 3 MEASUREMENTS synaptic morphology. ‣ Some forms of memory may not exist physically: When we talk about episodic memories, we are talking about immaterial experiences, that most likely arose because of brain activity. ‣ Memory is indirectly observed (problem of measurement): its existence is inferred from behaviour (e.g., a change in level of performance). ‣ The vast majority of memory research discussed in this lecture uses the memory measures we discuss in the next slides: recall, recognition, implicit tests, reaction time, source judgment.  16 BEHAVIOURAL MEASURES RECALL TESTS ‣ In a recall test, you explicitly ask the participant to generate a target memory. ‣ Free recall. You provide a global cue and participant reports all they can recollect. ‣ Examples: “Tell me all the words you remember from last week’s session.”; “Tell me all you remember what happened yesterday.” 4 MEASUREMENTS ‣ Cued recall. You provide a specific cue to target a specific memory. ‣ Examples: “What word was paired with the word banana in last week’s session?”; “In what town were you born?”; “What happened yesterday right after you woke up?” ‣ Free recall is always harder than cued recall.  17 BEHAVIOURAL MEASURES RECOGNITION TESTS ‣ A recognition test asks participants to decide whether an item presented to them during the test matches an item that they had studied before. ‣ Old/new recognition (or yes/no recognition). Participant is presented with a single test item and has to decide whether this item was presented during the study phase. ‣ 4 MEASUREMENTS Examples: “Was the word banana on the list you studied yesterday?”; “Have you seen this woman before?” ‣ Forced-choice recognition. You provide several alternative items, and participant has to decide which of them they have encountered before. ‣ Examples: “Which of the following words was on the list: (a) banana, (b) tomato, (c) cucumber”; “Select which of the four pictures you have seen before” ‣ Variants are often abbreviated like this: 2AFC (2-alternative-forced-choice test; i.e., there are 2 items to choose from), 4AFC, etc. ‣ Recognition is always easier than recall.  18 BEHAVIOURAL MEASURES IMPLICIT MEMORY TESTS ‣ Participants are not told to recollect a memory, rather, the existence of memory is measured by how well they perform on the test. Typical design and results of an implicit word-fragment completion task 4 MEASUREMENTS  19 BEHAVIOURAL MEASURES IMPLICIT MEMORY TESTS Study Phase Test Phase (e.g, measuring reaction time) 4 MEASUREMENTS “Can you put this in a bag?” “What could this be?” Typical outcome: items seen during the study phase (primed items) are recognized faster than items not seen. Reaction Time is often used in memory research to assess how difficult a memory task is, or how much internal processing is required for it.  20 BEHAVIOURAL MEASURES SOURCE JUDGEMENTS ‣ Source memory is memory for how you acquired a memory (from where/ whom you got the information), but not memory for the content. ‣ In a sense, this is an aspect of episodic memory. ‣ For some types of memory, our source memory is worse than memory for the content. For example, we usually are excellent at recognizing previously seen 4 MEASUREMENTS faces, but not good at recollection where we have seen the face before. ‣ Reality monitoring relates to source memory in that it is the ability to identify whether a memory stems from an actual event, or an imagined event (i.e., fantasy). MEASURING BRAIN FUNCTION 4  22 BRIEF OVERVIEW OF TECHNIQUES NEUROIMAGING TECHNIQUES: EEG ‣ All neuroimaging techniques measure some form of brain physiology (e.g., activity, connection 4 MEASURING THE BRAIN density between brain areas, etc.) during phases of a memory experiment to learn how performance in the memory experiment relates to these physiological measures. ‣ EEG (elecroencephalography). The oldest neuroimaging technique (1940s). Electrodes placed onto the scalp measure changes in the electrical activity of cortical neurons. Does not allow measuring deep brain structures like hippocampus, and the signal does not have high spatial resolution, but very good temporal resolution.  23 ELECTROENCEPHALOGRAPHY (EEG) METHOD AND OUTPUT OF EEG NAMES FOR ELECTRODE LOCATIONS SOME EXAMPLES OF EEG WAVES 4 MEASURING THE BRAIN  24 ELECTROENCEPHALOGRAPHY (EEG) NEUROIMAGING TECHNIQUES: EEG ‣ All neuroimaging techniques measure some form of brain physiology (e.g., activity, connection 4 MEASURING THE BRAIN density between brain areas, etc.) during phases of a memory experiment to learn how performance in the memory experiment relates to these physiological measures. ‣ EEG (elecroencephalography). The oldest neuroimaging technique (1940s). Electrodes placed onto the scalp measure changes in the electrical activity of cortical neurons. Does not allow measuring deep brain structures like hippocampus, and the signal does not have high spatial resolution, but very good temporal resolution. ‣ The high-temporal precision is evidenced in Event-Related Potentials (ERPs). ‣ ERPs are EEG responses related to specific repeated “events”. For example, in an experiment participants are presented several times with stimuli of different types (e.g., words and pictures). To calculate the ERP, the average EEG activity to the same stimulus type is averaged, which reduces noise.  25 ELECTROENCEPHALOGRAPHY (EEG) EXAMPLE USING ERPs IN MEMORY RESEARCH ‣ Participants were presented with a list 4 MEASURING THE BRAIN of words and then participated in a yes/no recognition task. ‣ Responses analyzed here: “Hit” means participant correctly recognized a word presented during the test as a word from the learning list. “Correct Rejection” means that participant The figure shows ERPs at electrode P3. Blue-shaded correctly identified that a word areas mark statistically significant differences. presented during the test was not on the learning list. ‣ For Hits, stronger negative deflection at 70-180 milliseconds before response is given. Also differences in the “late positive response) for both response types. Sun, Osth, & Feuerriegel2024 Cortex 176  26 MAGNETOENCEPHALOGRAPHY (MEG) MEASURING BRAIN ACTIVITY VIA STRENGTH OF MAGNETIC FIELDS ‣ MEG 4 MEASURING THE BRAIN (magnetoencephalography). A magnetic sensor detects changes in neuronal electric activity, expressed in strength of the magnetic field. This technique has high temporal resolution, and higher spatial resolution than EEG, although overall MEG and EEG have poor spatial resolution (unclear where activity is occurring).  27 USING MEG IN MEMORY RESEARCH COMPONENTS OF ITEM MEMORY RECOGNITION ‣ The topographical maps show magnetic field strength (fT=femto 4 MEASURING THE BRAIN Tesla) when subtracting strength for correct rejections (CR) from hits in a word recognition task. a) shows magnetic field average for 300-350 ms after word is presented. b) shows magnetic field average 350-400 ms after word is presented. ‣ Activity in a) linked to recognition of a word; activity in b) linked to memory for the context (background scene) in which the word was presented. ‣ In patients with damage to the hippocampus, the activity pattern shown in b) was not found. ‣ Overall, this research suggests that the hippocampus is not involved in word recognition, but in recognizing the context in which the word was encountered. Horner et al. 2012 Curr Biol 22  28 FMRI NEUROIMAGING TECHNIQUES ‣ fMRI (functional magnetic resonance imaging). Uses strong magnetic fields to align 4 MEASURING THE BRAIN hydrogen molecules (protons, abundant in the brain) to the field, disturbs their alignment with radio waves, and them measures the energy that is released when they re-align with the magnetic field. ‣ When neurons are more active, they require more oxygenated blood, and the fMRI can pick up the difference between oxygenated and deoxygenated blood. This is the Blood Oxygen Level- Dependent (BOLD) signal, which is thought to indirectly reflects neural activity. ‣ This method has relatively poor temporal resolution, but good spatial resolution. The stronger the magnetic field generated, the higher the spatial resolution: This method can detect changes to brain activity about 1-2 seconds after stimulus presentation, with activity peaking 5-6 seconds, and returning to baseline 10-20 seconds later. ‣ fMRI is expensive (one hour costs 500 CAD at McGill), and data analysis complex — makes it more unlikely that studies are replicated.  29 FMRI NEUROIMAGING TECHNIQUES ‣ Study design requires carefully selected conditions to compute differences in baseline brain activity 4 MEASURING THE BRAIN (e.g., looking at flowers) and activity related to target stimulus or stimulus-related processing (looking at human faces). ‣ Difficulty arises when to decide which difference in brain activity is considered meaningful. Usually, there are activity changes in most, if not all brain areas. ‣ Multi-voxel pattern analysis (MVPA). Using machine learning approaches to detect co- occurring patterns of activity in regions of interest that are related with specific stimuli or tasks. ‣ This allows to detect the overall brain response when a particular stimulus is presented or task is being performed (e.g., when looking at faces). ‣ This “blueprint pattern” can then be used to analyze fMRI data, to detect these patterns in brain activity. ‣ Has led to ideas that fMRI signals can be used to “read minds” — is this person lying or not? Outcome of these approaches controversially debated.  30 FMRI METHOD AND OUTPUT NEUROIMAGING TECHNIQUES: rFMRI 4 MEASURING THE BRAIN