Week 4 Flashcards
Libet’s experiment and wilful action
Wilful action usually involves an extended process of reflection, with feedback loops etc- This is not the case in this experiment
Can you predict behaviour better with fMRI and other neuroimaging methods than with psychological methods
NO
-Behaviour is hard to predict
-Many predictions are retrodictions
-AI developing algorithms to predict data- Threat to psychology
Prediction of behaviour (Bzdok & Ioannidis)
4 steps to predicting behaviour
1. Model building: Finding predictors for outcome- Choosing model that fits best- Significant correlation would mean somewhat accurate predictability
2. Internal validation: Test the model on data not used to build the model
3. External validation: Not done enough, Does the prediction work with external people
4. Generalisability and transposability- Exclusively WEIRD?
China using AI in primary schools
-3 electrodes sending real time data re brain waves to the teacher
-Details each students engagement levels at 10 minute intervals (Sent to parents)
-EEG v susceptible to artefacts
Bogus pipeline effect
-Common in forensic science
-Lie detection approach utilising a polygraph to measure variables in brain function/ skin conductance/ heart-rate/ breathing rate etc
-Usually doesnt detect lies but if someone is nervous
-People more likely to tell the truth
Cacioppo & Tassinary on Red
Paper about mind brain relationship
-Shows how in nearly all cases it is not a 1-1 relationship between psychological processes and physiological signs
fMRI experiment
Statistical models
-Univariate analysis: Compare experimental condition with data on a voxel-by-voxel (ROI) basis (Combining voxels to create ROIs improves signal: noise and multiple comparisons)
-Multivariate analysis: Correlating voxels with other voxels (Higher order correlation)- Not looking at one specific signal but looking at correlations/ relationships between signals
Searchlight strategy
-Take every voxel in the brain and compare them with each other
-Tried to predict when subjects were going to press left or right based on brain data prior to conscious awareness of the decision
-Based on data from 10s before decision (60% accurate)
-Brain areas associated with conscious control tasks were used
Neural correlates of sexual orientation
-Sexual content shown congruent or incongruent to their sexuality
-Subjective ratings of the images showed outliers and overlaps between groups
-Ventral striatum data matches pretty well to brain data
Interpretation of fMRI experiment
-Forward inference: During behavioural/ psychological task X engaging psychological process X, we found activation in brain area Y (Brain mapping)
-Reverse inference: During behavioural/ psychological test Z, we found activation in brain area Y, possibly engaging psychological process X (Dont know psychological process- Used for “mind reading”)- Dont know whether someone is lying or not, Need to infer from previous forward inferences
Conflict processing
-Associated with cingulate cortex and corpus callosum
-Idea that when people are lying, they are experiencing conflict which they have an intent of suppressing
Bayesian solution
Statistical equation showing probability of an inference
Delgado and brain chips
70s
Displayed in animals that he could control minds with the push of a button
Used electrodes on people who resisted previous treatments
Hess
Showed you can elicit behaviours by electrically stimulating different parts of the brain
Delgado and lobotomies
Thought lobotomies used by Moniz were too destructive and it was better to use electrical stimulation
Delgado and inducing emotions
Could induce emotions by stimulating different areas of the limbic system
-Triggering the septum could induce euphoria, counteracting depression and physical pain
Delgado taming a fighting bull
Caused a charging bull to stop a few feet before him by stimulating its caudate nucleus
Delgado and chimpanzees
-Programmed a stimoreceiver to detect spindles emitted by the amygdala
-When they were detected it produced an aversive reaction
-Within 6 days of this feedback the frequency of spindles dropped by 99%
-Calls this his most significant experiment
-Thought this could prevent disorders characterised by specific brain signals
Delgado and psychocivlized society
Claimed his technology was on the verge of creating a less cruel, happier and better man
Critics of Delgado
-Accused of trying to create a society where everyone who deviates from the norm will be surgically mutilated
-Also suggested brain stimulation is much less precise and beneficial than suggested
Delgado and non-invasive methods
Created a helmet that could deliver electromagnetic pulses to specific neural regions
-Success inducing drowsiness, alertness and treating tremors in Parkinsons patients
Brain implants today
-Cochlea implants most common today
-Clinical trials underway to stimulate vagus nerve to treat depression, OCD etc.
-Chips that stimulate optic nerve tested in blind subjects
-Electrodes that help with memory are a year or 2 away from testing in rats
(MI) Motor Imagery Data
Subject imagines the movement of a limb
What do BCIs do
Allow computers to be controlled by monitoring brain signal activity
-Imagining movements of a limb may be able to be used to control a BCI
Why are EEGs used for BCIs
High time resolution and cost effectiveness
-But prone to experience from external noise
Things that can affect EEGs
Posture of subject
Mood of subject
Duration of task- Poor concentration leads to poor EEG quality data
Evoked/exogenous EEG based BCI
-External stimulation required eg. visual stimuli
-These stimuli evoke responses that determine the will of the user
2 types of evoked BCI
- Dependent on visually evoked potentials- Brain signals generated in response to a viral stimulus
- Event-related potentials- Brain signals generated in response to sensory or cognitive events
Spontaneous/endogenous BCI
-No external stimulation
-Control actions are taken based on activity produced from mental activity
MI data may help in
Stroke rehab
Raw MI EEG data
Often averaged across numerous participants
-This hides higher or lower results
Pre-processing in MI EEG data
Aims to reduce the noise in EEG signals
-In turn increases the accuracy and robustness of BCI systems
Feature extraction in MI EEG based BCIs
Extracted features must show differentiation between task-specific brain states
Feature selection in MI EEG based BCIs
Detecting the most important data features in line with objectives of the study
-Simplifies the data set from the multiple input channels
Feature classification
Identifies the type of mental task carried out and activates the necessary control signals in the BCI system
Hybrid BCI
Combo of a BCI with another kind of interface
Application of BCIs
-Control: Manipulating an external device using brain signals
-Monitor: Identify the mental and emotional state of the user in order to control the environment they are in or the interface they are using
Uses of BCIs
-Replacement and restoration of CNS
-Therapy, rehab and assessment
-Affective computing
-Gaming
-Industry and transport
-Art
BCIs and replacement and restoration of CNS
-Paralysis, spinal cord injury, locked-in syndrome
-Difficult to develop due to altered brain functionality of these patients
-Could replace robotic prosthetics which rely on electrical signals in the muscles
-MI data can be difficult to generate in the brain
BCIs for therapy, rehab and assessment
-Stroke rehab can be aided by using BCI-controlled robotic arms
-Additional refinement is needed for systems to pass clinical trials
-VR trials have been used as people can control an avatars movements via their brain
BCIs and affective computing
-Users mood and psychological state are monitored to manipulate their environment to enhance or alter that state
-Possibility for monitoring to be used to push targeted marketing or political agendas- Ethical issue?
BCIs and gaming
-Currently a poor replacement for traditional methods
-Can be used to control difficulty in games as it measures the ability of the player and can increase and decrease the difficulty of the game
BCIs, industry and transport
-Could keep humans away from dangerous activities
-Could monitor when the user is too tired or unwell to carry out their job
BCIs and art- 4 types for artistic applications
- Passive: No active input from user who selects a pre-programmed response
- Selective: User has some restricted control over the system but doesn’t have a leading role in the creative output
- Direct: Select options from detailed menus to gain high levels of control
- Collaborative: Systems controlled by many users at once
Placement of BCI in ALS patient
Electrodes over motor cortex and transmitter on left side of thorax
How locked-in individuals communicate
Eye tracking
-When this fails they use eye movements/blinks in response to closed-ended questions (This limits options for independent communication- Associated with quality of life)
There is limitations of BCI for patients with
-Cortical damage
-Cognitive impairment
-Unsupportive caregiving
