Experimental Design

Experiments:

• subjects randomly assigned to groups

• composed of experimental group (this group is manipulated)

• composed of control group (being compared to experimental group)

• independent variables (manipulated)

• dependent variable (measured)

• provide evidence for causality (independent variable caused whatever change seen in the dependent variable)

Example:

• subjects randomly assigned in this experiment, different doses of MDMA given to subjects + placebo given (control group), to see its influence on the concentration of oxytocin in the blood.

Correlational Studies:

• 2 variables measured in the same subjects

• can be reffered to as predictor variable and outcome variable

• NO MANIPULATION of variables (there is no independent variable)

• DO NOT provide evidence for causality.

Example:

• height at 2 years old can predict that somebody will become a certain height in adulthood. there was no manipulation to make the kid grow.

• seeing the correlation between stress and cortisol level (positive correlation)

Designs:

• Testing it in the same person (within-subject design)

• Between subjects (not testing the same subject twice) (you’d wanna do this once because the independent variable is doing some permnanent effect) (for example, damaging a lesion of the brain to test its function)

Correlations:

• Positive: both measures increase or decrease together

• Negative: one measure goes up and the other goes down

• No correlation: no relationship

• non linear correlation:

Directionality problem

• physical exercise leads to more dendritic branching (don’t know cause)

3rd variable problem

• overall health could be causing dendritic branching or increase in learning scores

Spurious Correlation

• shark attacks go up during the summer, so does icecream sales. does the either cause the latter? no there is a 3rd variable (summer).

Determing causation when variables cannot be manipulated in humans:

• Converging operations: using multiple correlational, experiemntal, and quasi experimental approaches

  • Quasi experimental: covid study on respiration (one population had covid, one didn’t)

  • experiments in animal models or cell lines

  • studies in humans (epidemiological, longitudinal)

• Preponderance of the evidence: all the data points in the same direction

  • All the converging operations above suggest the same cause of a particular phenomenon

True experiments are not always possible! For example, converging operations was used to explain the causality between smoking and lung cancer. Another example is zika virus causing microcephaly.

Somatic & behavioral Vairables:

• somatic means the body

  • objective measure of brain anatomy or physiology

    • ex. hippocampul volume

• lesion

• drug injection

• hormone level

• electrical stimulation

• Behavioral variables

  • memory tasks

  • experiences (stressor)

Can be either IV or DV

Validity

• internal validity: is the cause and effect relatinoship due to the IV or are other factors be at play

• External validity: can this relationship be generalized to other populations