Please enable JavaScript.
Coggle requires JavaScript to display documents.
Thinking Critically About Neuroscience - Coggle Diagram
Thinking Critically About Neuroscience
Signal vs Noise
Separate Signal & Noise
Theory
Hypothesis
Experimental design
Appropriate analysis
Data sharing
Theory (Haslbeck et al., 2021)
Strong, falsifiable, make specific predictions
Qualities that make good theory
Phenomenon appropriately established
Why it caused should be able to be measured/operationalised & tested
Make specific predictions about human behaviour
Theory Construction & Use of Data Models
Generate = Abduction to establish phenomenon
Develop = Abduction to & deduction from to compare implied & empirical
Test = Deduction from to evaluate risky prediction
Causal Formal Model
Causal diagram to establish causal picture for formal theory
Code formal theory mathematically
Simulate data to predict behaviour using the code
Theories in Applied Settings
CBT
Prediction error
Prediction Error of Dopamine & Codeine (Schultz et al., 1997)
Relationship between surprise in environment & dopamine spike in striatum
No predicted but reward = Spike
Predicted & reward = Higher spike
Predicted but no reward = Depression in activity
So must show how surprise encoded in dopamine system
Cognitive Behavioural Therapy
Example of theory that makes particular predictions
How thoughts, emotions, behaviour & bodily sensations
Sessions follow general structure
Identify situation that trigger problem
How situation affects thoughts --> emotions --> behaviours --> bodily sensations
As interrelate therapist find what cause problem
Strategies to target area that cause problem
Theories
Social brain hypothesis
Model-based & model-free learning
Model-Based & Model-Free Learning (Russek et al., 2017)
If some map of environment = Expectations nested in cortical structure
Suggest systems encoded in diff regions
Model free = Striatum
Model-based = Prefrontal cortex
Social Brain Hypothesis (Dunbar & Shultz, 2007)
Humans developed neocortex to support more complex & dynamic social groups/interactions
Look cross-species
Relationship between neocortex & size of social group correlated
How linked to causal theory if true
Hypothesis
Closed science & pitfalls of HARKing
Avoids proliferation of ideas & effects which are subject to bias & unlikely to be reproduced
Must be derived from theory
Stronger theory = Better hypothesis
Not necessary for exploratory analysis
Open Science
Crucial step in development of strong theory & repeatable, replicable & reproducible effect
Features of Open Science
Open scientific knowledge
Open science infrastructures
Open engagement of social factors
Open delegation w. other knowledge systems
Replication Crisis (Open Science Collaboration, 2015)
100 replications from 270 authors
35 effects replicated
Experimental Design
Sample size & task
Should test theory & explain counterfactuals
No. Participants (Marek et al., 2022)
Power analysis simulaton
BWAS needs 1000s participants to be reliable
Low Sample Size (Szucas & Ionnidis, 2020)
Mist highly cited neuroimaging experiments have average sample size of 12
1000+ citations
300 studies 2017-2018 sample size 23-24
3% have power calculations
Power Calculation
Way to find if experimental design has no. people needed to reliably test
Analysis
Approaches to design
Can include variability in pre-processing, stat tests & reporting
Approached in many diff ways so essential to be transparent
Single Hypithesis (Botvinik-Nezer et al., 2020)
Single hypothesis run by 64 diff teams
Same hypothesis given to each team
Team chose how to conduct study
Yielded multitude outcomes
Data & Code Sharing
Improve way can replicate analysis
See how done & run same code to test if same findings
When comes to stating findings means can be open, clear & public
Understanding Psychology
How to think about theory
What is important about theory
Dead Salmon Experiment
Experiment (Bennet et al., 2009)
Dead salmon participated in sequential social task
Asked to determine emotions of series photographs to humans
Task
Shown series photographs depicting human individuals in social situations w. specified emo valence
Determine what emotion individual in photo experiencing\n- Stimuli presented in block design
15 photo presented for 10s w. 12s rest
Findings
Several active voxels discovered in cluster in brain cavity
Size cluster = 81 mm3
Cluster-level significance = p=.001
Identical t-contrasts controlling false discovery rate & familywise error rate completed
Indicated no active voxels even @ relaxed stat thresholds p=.25
Explanation
Human error
Strange stat error
Stat imprints try to find patterns in data
Significance
Can find findings in seemingly impossible
Depression & the Brain
Experiment (Daws et al., 2020)
Psilocybin a psychadelic
Psilocybin shown to act as potential antidepressant
But mechanisms of action unknown
Study
Participants w. treatment resistant depression
Measured w. Beck depression inventory
Open label phase I trial of psilocybin (10mg & 25mg)
Double blind phase II trail of psilocybin & escitalopram (10-20mg)
Had registered QIDS & primary outcome measure
fMRI used on both trials
Baseline & 3 weeks after use
Issue
Can be obvious if given active compound
Findings
QIDS not significant between groups
Pre-registered questionnaire showed comparable reductions in depression for both medications
Used Beck depression inventory as key finding
Key Exploratory Theory
High entropy --> Repetitive - thoughts --> Depression
Entropy = Thoughts as more noise in brain
Thoughts = Depression as ingrained thoughts
So can decrease depression by decreasing entropy
Issues\
Psychological treatment
Unblinding
Obvious if active/inactive compound
Spurious stat assessment
No stat comparisons
Found something that worked & reported as if it was primary aim
Theoretical assumptions
Reverse inference
Look at something in brain & argue must be related to what trying to study