Lecture 4 - Relationships and Inference

Question 1

What does a one-to-one (isomorphic) relationship mean?

Also give examples;

Answer 1

“I smile when I am happy”

• I always smile when I am happy
• I am always happy when I small

–> This rarely ever happens

Question 2

What does a many-to-one relationship mean? Also give an example;

Answer 2

“I smile when I am happy”

• I smile when I am happy
• I am not always happy when I smile
• there are other reasons why I might smile

Question 3

What is a common mistake for specificity, and what does it imply?

Answer 3

Affirmation of the consequent
= assuming that if we know what psychological event gives rise to a physiological response, that we also assume the opposite

So when I am nervous I sweat, does not mean that when I sweat, I am nervous. Might be other reasons for sweating too

Question 4

What does generality mean in relationships between psychology and physiology?

Answer 4

Relationship between psychology and physiology can be context dependent or cross-situational

Question 5

What are the four (/five) relationships of psychology and physiology?

Answer 5

1. Outcome - Not specific and not generalizable
2. Concomitant - Generalizable (can happen in both ways) but not specific
3. Marker - Specific but not generalizable
4. Invariant - (Holy grail!) - Both specific and generalizable; I am always happy when I smile and it does not matter when or where
5. Null relationship - No relation between a certain combination of psychological and physiological events

Question 6

How to correct for interpersonal differences in reactivity and range?

Answer 6

Use baseline task:

• Feature scaling (values are expressed as a fraction of the range)
• Z-scores (values expressed as number of standard deviations from the mean)

Question 7

When do you not want to normalize?

Answer 7

• When you are interested in interpersonal differences
• When you have a between-subjects design; since normalizing may mask the effect of the between-subjects manipulation
• Within-subject designs are therefore preferable, however this is not always possible

Question 8

How to represent a normal level and a normal range

Answer 8

Normal level with baseline

Normal range with reactivity

Question 9

How to deal with confounds? Two ways:

Answer 9

1. Control for them in your experimental design (this is the best option, but it is hard to control for everything in a study)
2. Control for them in your analysis