Modern Research Methods: Introduction 13 January 2020 Molly Lewis
Dr. Molly Lewis • Psychologist interested in understanding how languages are learned and change over time • PhD from Stanford University in Developmental Psychology • In my own research, use combination of classical experimental methods and large scale data • Care a lot about open science and tools for open science
TA: Jaeah Kim • 4th year PhD student in the Psychology department. • Did my undergrad at Carnegie Mellon University as well. • Currently interested in seeing how we can measure attentional states from eye-tracking data and studying how some attentional abilities might develop over childhood (especially between ages 3 and 5).
1. 1. Ob Observatio ion The Scientific Process 2. 2. 3. 3. … n . Scie ientif ific ic theory THEORY
Case Study of the Scientific Process ….in Developmental Psychology
A guy jumps in the swimming pool with all his clothes on. Why?
“Theory of mind” (ToM) • Maybe someone was drowning? • Maybe he saw a $20 bill at the bottom? • Maybe he was drunk and thought it would be fun? • Having a “theory of mind” allows you to reason about the guy’s behavior when he jumped in the swimming pool • He had something in his head (a belief) that caused him to do what he did
The origins of “theory of mind” In assuming that other individuals want, think, believe, and the like, one infers states that are not directly observable and one uses these states… to predict the behavior of others as well as one’s own. These inferences, which amount to a th theory ry o of f min mind , are, to our knowledge, universal in human adults. Premack & Woodruff (1978): “Does the chimpanzee have a “theory of mind”?
False belief: a key test Child knowledge False belief consistent consistent
The Sally-Anne Task • Failure at 3 years, success at 4 • Developmental pattern is very robust • Wh Why do children fa fail? • Demands of the task: • Represent the true state of the world • Represent false belief • Attribute false belief to another person • Select between them on the basis of a linguistic prompt Wimmer & Perner (1983)
Meta-analysis of ToM tasks … Things that mattered: “Anne” having a motive • Child’s participation • Physical presence of object • Salience of mental state • Wellman et al. (2001)
Related achievements • Ability to imagine alternate realities (pretending) • Ability to make people believe things that are false (lying) • Ability to take different perspectives on the same scene (representation change) • Understanding appearances can be deceiving (appearance-reality distinction)
Social cognition in other primates Hermann et al. (2007)
A beautiful story • Children begin reasoning about agents’ desires, goals, and actions • They then gradually develop a representational theory of mind around 3 years of age • Theory of mind is only seen in humans … • But what if infants could also represent others’ beliefs?
Evidence for early theory of mind? 25+ papers from 10+ different labs (e.g. Buttelman et al, 2009; Clements & Perner, 1994; (Closes lid) Knudsen & Liszkowski, 2012; Onishi & Baillargeon, 2005; Southgate et al., 2007; Southgate et al., 2010 ) (Where will the (Secretly removes lady look?) ball) 17/20 24-month-olds looked at the correct (belief-consistent) window (right) Southgate et al. (2007)
How do we resolve this discrepancy? • Collect more data - Are we sure this pattern is correct? https://manybabies.github.io/ • Revise the theory • Complete continuity • Preschool results are artifacts • Standard tasks too difficult • TOM 1 and TOM 2 • Implicit system and explicit system • One early/innate, a second one learned slowly • Other possibilities?
The Scientific Process THEORY 1 662 Child Development only what we termed primary conditions. These were A 1 1.01 conditions in which (1) subjects were within 14 1T, oo .9 months of each other in age, (2) less than 20% of the o• .8 0 initially tested subjects were dropped from the re- 00 ported data analyses (due to inattention, experimen- tal error, or failing control tasks), and (3) more than 6 .6 u o8 80% of the subjects passed memory and/or reality ao o, o ) .5 o0 control questions (e.g., "Where did Maxi put the chocolate?" or "Where is the chocolate now?"). Our .4 O reasoning was that age trends are best interpretable if o 3 each condition's mean age represents a relatively nar- .2 0 o's( o row band of ages; interpretation of answers to the tar- 000W get false-belief question is unclear if a child cannot re- .1 o -- THEORY 2 member key information, does not know where the 30 40 50 60 70 80 90 100 110 object really is, or cannot demonstrate the verbal facil- ity needed to answer parallel control questions. In Age (Months) most of the studies, few subjects were dropped, very high proportions passed the control questions, and B S00 ages spanned a year or less, so primary conditions in- 0 0 1 cluded 479 (81%) of the total 591 conditions available. D oa o 5o ..I . /1oo The primary conditions are enumerated in Table 1; 4 - they were compiled from 68 articles that contained 0 128 separately reported studies. Of the 479 primary /-0 0 3 00 o o! /O/ conditions, 362 asked the child to judge someone 2 0 0oo /( a 04 o 4 - 2%c o 1 o _0o else's false belief; we began our analyses by concen- 70 O 00 '-o! o o o trating on these conditions. On average in the pri- 0 of children were dropped from a mary conditions, 3% condition, children were 98% correct on control ques- tions, and ages ranged 10 months around their mean 2 values. 00-3 00 In an initial analysis only age was considered as a 0 factor. As shown in Figure 2, false-belief performance -4- 000 dramatically improves with age. Figure 2A shows each primary condition and the curve that best fits the 30 40 50 60 70 80 90 100 110 data. The curve plotted represents the probability of Age (Months) being correct at any age. At 30 months, the youngest age at which data were obtained, children are more Figure 2 Scatterplot of conditions with increasing age show- than 80% incorrect. At 44 months, children are 50% ing best-fit line. (A) raw scatterplot with log fit; (B) proportion correct, and after that, children become increasingly correct versus age with linear fit. In (A), each condition is rep- correct. Figure 2B shows the same data, but in this resented by its mean proportion correct. In (B), those scores are case the dependent variable, proportion correct, is transformed as indicated in the text. transformed via a logit transformation. The formula for the logit is: straightened, yielding a linear relation that allows , logit = In systematic examination of the data via linear regres- sion; second, the restricted range inherent to propor- tion data is eliminated, for logits can range from where "ln" is the natural logarithm, and "p" is the negative infinity to positive infinity; and third, the proportion correct. With this transformation, 0 rep- transformation yields a dependent variable and a resents random responding, or even odds of predict- measure of effect size that is easily interpretable in ing the correct answer versus the incorrect answer. terms of odds and odds ratios (see, e.g., Hosmer & (When the odds are even, or 1, the log of 1 is 0, so the logit is 0.) Use of this transformation has three major Lemeshow, 1989). The top line of Table 2 summarizes the initial anal- benefits. First, as is evident in Figure 2B, the curvilin- ysis of age alone in relation to correct performance ear relation between age and proportion correct is
The Scientific Process is cumu mulati tive
Cumulative science is hard Sometimes doesn’t always work the way it should…. This class is about learning how to think of ps psychol ology ogy as a ce , and learning pr ods for doing so. cu cumulative ve sci cience practical me methods
Overview of course 1) Philosophy of Cumulati 1) tive Science 2) 2) The Single Experi riment t – Experimental design, tools in R for working with data and plotting data, reproducibility t – Intro to statistical 3) 3) Repeati ting an Experi riment concepts, replication of experiments 4) 4) Aggregati ting Many Experi riments ts – Meta-analysis
Course Logistics Website: https://cumulativescience.netlify.com/ Includes schedule and readings, contact info, grading and other policies Please read carefully, and let us know if you have questions.
Course Components • Lecture MW; Lab F (Porter 332P) • Some lecturing, some interactive • Encouraged to bring your laptop to lab • 8 assignments • Typically handed out in lab, and due Thursday at noon • Must be completed individually, but can work with others • Focus on R using Rstudio.cloud (more in lab on Friday) • Lowest score dropped • Take home midterm (completed individually) • Final project: Meta-analysis (completed in teams) • Participation
Recommend
More recommend
