Lab G - Neuro Research Methods

How to create a research question

• observe phenomenon

• ID gap in knowledge

• review what’s already known

• consider how gap in knowledge can be studied in a measurable way

• predict outcome/form hypothesis (BE SPECIFIC)

  • hypotheses support outcomes, do not PROVE them

  • make sure hypothesis can actually be demonstrated/rejected

How to design an experiment

• decide independent/dependent variables

• ID constants to avoid confounding influences

• goals: minimize error, allow for replication, produce results that reflect the effect of the variable being studied

independent variable

variable that’s intentionally changed/manipulated to test its effect

dependent variable

variable that’s measured or observed to determine the outcome of the manipulation

control variable

factor that’s kept constant across groups (ensures differences in results are caused by the independent variable)

confound

unintended variable that changes alongside the independent variable (could provide an alternate explanation for the results)

• makes it impossible to know which factor actually caused the observed effect

Choosing a sample

• goal: sample represents population

• random selection

• consistent handling

variation

natural differences between individuals within a population (e.g. genetics, experiences)

How to separate real treatment effects from random variation

• replication of study

• random assignment of participants

• statistical analysis

How to choose your sample size

via power analysis

power analysis

statistical calculation that estimates how many animals are needed to detect a meaningful difference between groups with a given level of confidence

• considers expected size of the effect

• considers variability within data

• considers desired probability of correctly detecting a true effect

randomization

each animal has an equal chance of receiving any of the treatment

replication

• repeating the experiment

• using multiple animals in each group

random number table

• list of randomly generated numbers

• numbers are completely independent of each other

• used to assign treatments to animals

animal models

allow neuroscience research that can’t be achieved in humans

• induced disease

• spontaneous changes

• genetically modified animals

• negative model

• healthy animals

induced disease models

• allow researchers to reproduce symptoms or biological changes similar to those seen in humans

  • pharmacologically induced

  • lesion induced

  • stress induced

  • biologically induced

pharmacologically induced disease model

administration of a chemical/toxin to alter neural activity or damage pathways

lesion induced disease models

physically or chemically damaging a brain region

stress-induced disease models

use chronic or unpredictable stress exposure to activate the HPA axis and study anxiety/depression-like behaviors

biologically induced disease models

introducing infectious agents/biological molecules (e.g. injecting amyloid plaques in healthy specimen to study Alzheimer’s)

spontaneous change models

rely on natural biological changes that occur in animals without any external manipulation (develop diseases/traits that resemble human conditions as a part of their normal life course)

genetically modified animal models

created by adding/removing/changing particular genes in order to study their role in normal function or disease

• knockout model = gene removed

• knockin model = modified or foreign gene

negative models

animals that are resistant or non-susceptible to diseases/conditions that affect other species

healthy animal models

establish baseline data on normal neural structure, function, and behavior

experimental design: between subjects

different groups of animals are assigned to different treatment conditions

experimental design: within subjects

uses the same animals across multiple conditions

experimental design: factorial

tests two or more variables at the same time to observe their individual and interactive effects

experimental design: repeated measures

measures the same animals at multiple time points (only one condition)

experimental design: mixed

combines elements of both between and within subject approaches

• different treatment between groups

• give treatment at different times within groups

experimental design: randomized block

aka stratified design; ID factor that could affect results, assigns animals from each factor group into every treatment

experimental design: latin square

controls for two sources of variation

• structured version of a within groups design - the same animal receives all treatments under different days/conditions

dose response

how the physiological/behavioral effect changes as the dose of the treatment changes

linear dose response

effect increases in direct proportion to the dose

physiological ceiling

limit to how high a response can go (similar to carrying capacity)

quadratic dose response

effect increases up to a point, then levels off or declines

cubic dose response

rise and falls in effect as dose increases

threshold

lowest dose required to produce a measurable effect

continuous data

numerical values that take on any value within a range

categorical data

distinct groups or types

ranked data

ranks have an order, but the difference between rank 1 and 2 may not be the same as the difference between rank 4 and 5

t-test

used when comparing 2 treatments/conditions (number of groups doesn’t matter)

vehicle

inert carrier substance used to deliver the treatment

standard error bars

predict how much the sample mean is expected to vary from the true population mean

Why are multiple t-tests not used when there are more than 2 treatments/conditions?

doing multiple t-tests increases the chance of a false positive (type 1 error)

ANOVA

• used when comparing more than 2 treatments/conditions (number of groups doesn’t matter)

• tells you if any of the treatment groups differ from each other (does not tell you WHICH)

Tukey’s test

tells you WHICH treatment groups differ from each other

p-value

• how likely is it that the difference between the treatment groups happened by chance?

• p-value < .05 —> difference is not random (is statistically significant)

one-tailed t-test

used when you’re testing for a difference in one specific direction

• e.g. a drug will increase food intake compared to control

two-tailed t-test

used when you’re testing for any difference in either direction

• a drug will change food intake compared to control

paired t-test (type 1)

used when the same subjects are measured twice

• e.g. before and after a treatment

two sample equal variance t-test (type 2)

used when comparing two independent groups that are expected to have similar variability

two sample unequal variance t-test (type 3)

used when comparing two independent groups that may have different variability

y-axis

• anterior-posterior (front-back)

• coronal plane

x-axis

• lateral adjustment (side to side)

• sagittal plane

z-axis

• depth adjustment (up and down)

• horizontal plane

cannula holder

holds cannula

ear bars

stabilize head by fitting into auditory meatus

incisor bar

supports upper incisors, levels skull so that bregma and lambda are on the same plane

nose clamp

holds nose in place

How to read a vernier

• read left side first (compare with 0 mark)

• then read right side (where lines line up)

Steps to stereotaxic surgery

• insert ear bars (make sure they read the same number)

• level bregma and lambda (z axis)

• zero the instrument (measure x,y,z of bregma)

REMEMBER THESE ARE IN CM