2: Theories and Questions Flashcards
what is the distortion rule
Procedures used to make our observations should not introduce distortions.
Sources of distortions
- From instruments (apparatus) used for measurements
- From observer / experimenter (observer bias)
- From sampling procedures
- From the environment (more so if not controlled, i.e.,
outside the laboratory)
when and how does observer bias occur
- During the experimentation or observation: giving clues to the subject, misinterpreting behaviours.
- During recording or analysis of the behaviour/ data (especially during times of uncertainty).
- Often unconscious
11 sacred principles of science
replication, speculation, hypothesis, data, questions, Paradigms, theories, models, principles, rules, laws & hypotheses
replication
- The basic idea of scientific experimental research is replication (to judge reliability), not statistical significance.
- Relevant to n-of-1 or case study research.
- Claude Bernard (1865): “Introduction à l’étude de la médecine expérimentale”.
- Experimentation = provoked observation.
- “Show me the data!”
Speculation, hypothesis, data
- Idiographic (the specific) or nomothetic (the general)?
- “The plural of anecdote is data” (Ray Wolfinger, circa
1969-1970). - Datum vs. data: Anecdotes / n of 1 observations have no scientific value?
- Certainly debatable in neuroscience, ethology, clinical neuroscience or psychology… and most purely idiographic sciences (e.g., astronomy).
what are paradigms
Set of laws, theories, methods and applications that
form a scientific research tradition
what are theories
A collection of hypotheses about a specific phenomenon. A set of assumptions about the causes of a behaviour and the rules that specify how the causes operate. Metaphor: Theory ≈ Map
what is a model
a specific implementation of a theory
what is a principle
a generally accepted ‘fact,’ not always tested
what is a rule
a generally accepted process or pattern, sometimes mathematically defined
what is a law
substantially verified theory
what is a hypothesis
Statement used to test a theory or model. A testable statement about the relation between variables. Paradigms, theories, models, principles, rules, laws & hypotheses
what are qualitative theories
verbal statements, discourse-
based. Variables can be discussed, but are not necessarily mathematically evaluated.
what are quantitative theories
Mathematically / statistically
inspired. Relationship between variables and constants are investigated. Rules, formulas,
computational models are used
what are descriptive theories
Describe relationships between variables, no explanations given
what are analogical theories
The relationships between
variables are explained via analogies or metaphors
what are fundamental theories
Complex new constructs
and concepts are suggested
domain or scope of theories
theory of evolution
learning theory
neural network theory
super-male (testosterone) theory of autism
what is the theory of evolution
broad scope for biology (evolutionary biology), neuroscience (comparative neuroscience, neuroethology, neurobiology), psychology (evolutionary psychology, animal psychology), anthropology (biological anthropology, primatology), etc
what is learning theory
skinnerian principles
other examples of theories
- Information theory // Communication theory
- Decision theory / Signal detection theory
- Psychoanalytical theories (Freud, Jung, Adler)
- Attachment theory (Bowlby, Ainsworth, etc.)
- The “Theory of Mind” theory
- Triune Brain theory (MacLean)
- Hebbian theory (Hebb’s rule or Cell Assembly Theory)
- Ramachandran // Michael Persinger: Neurotheology?
roles of theory in science
- Describing phenomena
- Understanding phenomena: Finding the cause
- Predicting phenomena
- Explaining phenomena: Organizing and interpreting research results
- Generating research: Heuristic value of theories.
a good theory…
- Can account for the data collected.
- Has explanatory relevance (logical soundness).
- Is testable: Can be verified / confirmed or disconfirmed (i.e., falsifiable)*. Testable = falsifiable.
- Predicts novel events. Prediction
- Is parsimonious. Parsimony
- No “proof” in science (except mathematics?), only plausibility
steps in developing theories
- Defining the scope of the theory
- Reviewing the literature
- Formulating the theory
- Establishing predictive validity
- Testing the theory empirically
hypothetical construct
Cronbach and Meehl (1948), McCorquodale and Meehl (1948)
- Inferred, but untested. Not operationally defined.
- Cannot be observed directly.
- Properties and implications not demonstrated in empirical research.
- “Intelligence”, “motivation”, “personality”, “stress”, “emotions”, “moods”
intervening variables
From Tolman (1938)
- Hypothetical internal, covert, implicit state
- Inferred, summary of empirical data; operationally defined.
- Cannot be observed directly, at least, not initially: “valence” (Lewin), “habit strength” (Hull), etc.
- The variable used to address an hypothetical construct
hypotheses
- “Asking questions” = “formulating hypotheses”.
- Experimental psychology and neuroscience being mainly an hypothetico-deductive sciences, hypotheses are very central to the research process.
- Hypothesis: Tentative explanation.
- The tentative explanation often includes a statement about the relationship between two or more variables.
- The hypothesis should be testable.
steps in Experimental Research
- Ask a question: from data, observations, theories, etc. Develop a research idea.
- Make preliminary observations (pilot project) and start to formulate hypotheses.
- Make predictions from hypotheses that can be easily tested empirically.
- Identify the variables that need to be measured; define the problem: give operational and/or ostensive definitions (defined later).
- Select a research approach (correlational, observational, experimental, hybrid) and design.
- Select subjects / participants / instances / model system: human participants, animal subjects (what species?), cell lines, etc.
- Observations and measurements / measures: Choose a suitable research environment, necessary equipment, and choose the right sampling, recording & scoring methods, etc.
- Collect sufficient data and make sure you have enough subjects / participants / observations in order to validate or invalidate your hypotheses.
- Use the appropriate statistical data analysis (or analyses): exploratory or confirmatory data analysis; descriptive and/or inferential data analyses.
10.Report your results / dissemination: Talks, posters, papers, book chapters, etc
what are some alternative approaches
- Bayesian approach: Prior probabilities (as opposed to the “frequentist” approach [Neyman–Pearson]). The “new
inductivism”. - Strong inference (J.R. Platt, 1964): Inductive inference;
more below. - New experimentalism (Ackermann,1989): A-theoretical. For example:
* Severe experimentalism
(Deborah Mayo, 1996);
more below.- New statistics (Cumming,
2012): Estimation thinking;
Effect sizes and CI’s.
- New statistics (Cumming,
strong inference, who suggested it, process of elimination, works well with…
- Strong inference was suggested by Platt (1964):
- Process of elimination: Develop several alternative explanations (leading to testable predictions) and test all the predictions.
- Works well with high control in an experimental context. For example, popular in molecular biology.
- Works well with precise predictions (i.e., clear precise outcomes).
strong inference strategy
- Devise alternative hypotheses (note the plural).
- Devise an experiment (or experiments), with alternative possible outcomes. Ideally, these outcomes will help you reject one or more of your
hypotheses. - Carry-out the experiment(s).
- Repeat the procedure, devising sub-hypotheses or sequential hypotheses in order to define and refine the possibilities that remain.
Strong inference strategy quote from platt
quote from platt (1964)
1) Devising alternative hypotheses;
2) Devising a crucial experiment (or several of them), with alternative possible outcomes, each of which will, as nearly as possible, exclude one or more of the hypotheses;
3) Carrying out the experiment so as to get a clean result;
1’) Recycling the procedure, making sub-hypotheses or sequential hypotheses to refine the possibilities that remain; and so on.
what does strong inference reduce the chances of
confirmation bias:
* Tendency to look for confirmatory information and ignore contradictory information.
strong inference has a method to devise alternative hypotheses: what two preliminary phases are used before starting the experiment
- Exploratory phase: preliminary observations, tests
- Pilot phase: mini-experiment (with small sample)
Severe experimental testing (Mayo), error, progress, and strategies
Error detection and error correction
Progress from the identification of errors (learning from our mistakes): Induction as a self-correcting process
Strategies:
* Practical interventions
* Cross-checking
* Error control and elimination, error statistics
