Legal Issues

Question 1

1. Why are neural correlates still informative about mechanisms? What are the limitations of correlates?

Answer 1

Limitation
• Many psychological phenomena, like personality, can only be explored on observation basis
o We know which personality trait correlates with which brain variables etc.
• Correlation is not causality! We do not know if a third variable plays a role
• Experiments allow us to directly relate elicited neural activity to the psychological process – we do not know, however, that the observed brain activity is causal for the process

Still useful!
• Nice starting point for forming theories
• Theories can be strengthened by including possible confounds/covariates

Question 2

2. Which neuroscience methods provide causal and which correlational evidence?

Answer 2

Correlational: fMRI, EEG, MEG, Single-Cell recording.
Causal: TMS, lesion studies
(For more, look at Steffi’s Notes).

Question 3

3. What are standard fMRI analysis, MVPA and Representational analysis used for?

Answer 3

fMRI: brain mapping – linking mental processes to regions they activate
• Brought us knowledge about associations between function and structure
• But not very helpful in understanding how psychological functions are implemented

MVPA: brain-reading – decoding psychological states from patterns of brain activity
• Helps understanding how cognitive functions are organised in the brain

Representation similarity analysis (RSA)
• Asks how patterns of brain activity evoked by different stimuli are related  how are mental representations implemented in the brain?
• Allows to test
o theories on similarity between objects, like categories
o Similarities across species, like humans and non-human primates

Limitations
• Difficult interpretation of multivariate approaches
• MVPA is limited by spatial characteristics of neuronal representations
• Again: no causality inferable

Question 4

4. Should neuroscience be used to predict personal characteristics, such as health, academic achievement, and criminal behaviour? Should brain enhancement be allowed? What are the dangers of this use of neuroscience?

Answer 4

Neuroscientific predictions
• Use must be managed carefully  protect privacy and avoid discrimination
• Predictions are generalisations based on groups  not reliably applicable to individuals
o Public should be protected from claims that do not hold, e.g. expecting a fully scientific way to detect lies

Brain enhancement
• Fairness – is it like doping in sports? What about the gap between rich and poor?
• Social standards – will our normal performance become sub-standard?
• We do not know much about the mechanisms of e.g. neurostimulation  safety and efficacy need to be established

Question 5

5. Problems and solutions to reproducibility problem.

Answer 5

Why is reproducibility especially problematic in neuroimaging?
• Data’s high dimensionality
• Relatively low statistical power in many studies
• High degree of flexibility in data analysis
• Potential for questionable research practices, like circular analysis

Potential solutions
• Data sharing projects
• Free and open source software packages
• Publication of reproducible analysis workflows
• Machine learning methods, focusing on generalisation, rather than on statistical significance
• To avoid circular analyses: Predictions need to be developed and tested on different population, to avoid “optimism bias”