Brain & Cognition – Revision Lecture Vocabulary
General Orientation
Purpose of session: rapid revision of critical / high-difficulty topics for upcoming MCQ exam (≈ next Friday).
Material today is NOT sufficient alone; study all spring-term lectures, readings, slides & recommended papers.
Neural Networks & Machine Learning (Amir)
Core vocabulary
Node / Neuron: decision unit; receives inputs, produces single output.
Layer types: input, hidden, output. Multi-layer collection = network.
Weight (): strength of influence along a connection; multiplies incoming signal.
Bias (): constant added before activation; shifts decision boundary.
Activation function (): maps summed input to output (e.g. step, ReLU, linear).
Perceptron: minimal NN (1 input layer ➜ 1 hidden layer ➜ 1 output).
Training paradigms
Supervised: labelled data (e.g., MRI scans labelled "Alzheimer" vs "healthy", cat–dog images).
Unsupervised: discover structure w/o labels (Amazon & YouTube recommenders).
Reinforcement: trial-and-error, reward-driven (self-parking car; Atari paddle game; mouse in maze).
Testing / validation = evaluate on novel set distinct from training.
Typical MCQ style
Given , compute network output and compare to threshold.
Interpret graphs of activation functions (e.g., ReLU outputs for ; linear elsewhere).
Concept links
Strength analogy: weight ≈ synapse efficacy; bias ≈ neuron excitability.
Hidden layers learn intermediate representations.
Language, Communication & Pragmatics (Ether)
Language in real life = dialogue, not isolated monologue. Requires modelling listener’s mental states.
Pragmatics vs semantics & lexicon: meaning shaped by context, speaker identity, shared knowledge.
Interactive Alignment Model (Pickering & Garrod)
Speaker & listener align at multiple linguistic levels (phonology ➜ syntax ➜ semantics).
Alignment cascades, enabling rapid prediction of upcoming utterances.
Gricean Cooperative Principle
Maxims: Quantity, Quality, Relevance, Manner.
Listeners assume cooperation; violations trigger implicatures (sarcasm, hidden motives).
Experimental evidence
ERP shows word comprehension modulated by speaker gender/class.
Eye-tracking: shorter descriptions with familiar partner; adaptation to listener needs.
Invisible meanings: sarcasm conveyed via prosody, facial cues.
Social motivation
Older adults recruit extra resources when topic / partner is personally relevant.
Double-Empathy Problem (autism): mis-communication stems from neurotype mismatch, not one-sided ToM deficit.
Higher-order language phenomena
Counterfactuals: maintain parallel mental model & mark it as “non-true”. Shares ToM network; impaired in autism.
Figurative language: familiar metaphors understood directly (not 2-step); irony relies on context & attitude.
Face Perception (Marcus)
Faces signal identity, emotion, gaze, age, sex; identity often primary.
Familiar faces
Brain stores multiple instances ➜ constructs statistical average.
Mean represents stable traits; variance encodes appearance boundaries (fatigue, lighting).
Unfamiliar faces
High within-person variability → poor matching from CCTV, passports.
Individual differences: most people near chance; a few “super-recognisers”.
Professional performance
Passport officers ≈ novices; trained police facial examiners somewhat better.
Methodological innovation
Virtual Reality: avatar generation of real people; tests recognition & matching in controlled yet ecological scenes.
Psychologists validate avatars against photos (similarity metrics; behavioural performance).
Key computations: mathematical mean, variance, Euclidean distances in face space.
Cognitive Control (Sarah)
Controlled vs Automatic processing
Control needed when tasks are novel, conflicting, error-prone, or require inhibiting dominance.
Experimental paradigms
Stroop, Flanker, Go/No-Go, N-back, task-switching, dual-task, Stop-signal.
Classic theoretical frameworks
Central Executive (Baddeley & Hitch working memory).
Supervisory Attentional System (Norman & Shallice).
Brain substrates
Prefrontal cortex (PFC): dorsolateral (cold cognition, manipulation), ventrolateral (maintenance), medial & orbitofrontal (evaluation, motivation).
Damage ➜ “dysexecutive syndrome” (disinhibition, perseveration).
Key distinctions
Top-down vs Bottom-up control.
Amplification vs Inhibition of competing representations.
Emerging view: multiple large-scale networks (fronto-parietal, cingulo-opercular, default) flexibly couple to meet demands.
Link to intelligence (fluid ) but peripheral for exam.
Vestibular System (Maria)
Anatomy & basic functions
Two labyrinths (inner ears) each contain: 3 semicircular canals (detect angular acceleration) + 2 otolith organs (utricle/saccule; detect linear acceleration & gravity).
Primary pathways: labyrinth ➜ vestibular nerve ➜ vestibular nuclei & cerebellum ➜ (i) spinal cord, (ii) extra-ocular muscles, (iii) thalamus ➜ cortex (parietal operculum OP2) & hippocampus.
No unimodal vestibular cortex; signals widely distributed – especially parietal.
Reflexes
Vestibulo-Ocular Reflex (VOR): stabilises gaze during head motion.
Vestibulo-Spinal Reflex (VSR): maintains body posture/centre of gravity.
Vestibulo-Collic Reflex (VCR): stabilises head (bird-head video).
Research methodologies
Turntables – controlled, natural motion; passive movement.
Caloric Vestibular Stimulation (CVS) – warm/cold water in ear; clinical standard; short duration; side-effects.
Galvanic Vestibular Stimulation (GVS) – trans-mastoid current; remote-control sway; high parametric control; screening needed.
Cognitive & perceptual roles
Body perception: vestibular + gravity cues shape body schema, ownership, self-location.
Self-motion perception: visual–vestibular integration; conflicts (e.g., VR) → motion sickness.
Gravity processing: verticality, joint-position bias, exploration–exploitation balance; disrupted in micro-gravity → space motion sickness, re-adaptation issues.
Clinical conditions
Peripheral (labyrinth/nerve):
BPPV, Ménière’s disease, Vestibular neuritis – vertigo, oscillopsia, instability.
Central (brain): vestibular migraine, stroke – broader cognitive & orientational deficits.
Depression & Schizophrenia (Andrew)
Overarching model
Biopsychosocial: biological, psychological & social factors interact.
Major Depressive Disorder (MDD)
DSM criteria: persistent low mood/pleasure + associated symptoms ≥ weeks causing dysfunction.
Heritability: (family & twin studies).
GWAS: highly polygenic; > M individuals; implicates glutamatergic & GABAergic genes; no single “depression gene”.
Polygenic Risk Scores (PRS): predict small variance in cognition (mental flexibility).
Cognitive theories: Beck’s Negative Triad (self, world, future), attentional biases, negative schemas.
Imaging: hyper-activation ventral PFC (self-referential), hypo-activation dorsolateral PFC (cognitive control).
Schizophrenia
DSM: delusions, hallucinations, disorganised speech/behaviour, negative symptoms; 6-month disturbance.
Heritability: higher than MDD; still polygenic.
PRS associate with cognition & other psychopathology.
Generalised cognitive impairment (attention, memory, EF) independent of positive symptoms.
Predictive-Processing Model: imbalance of priors vs sensory evidence leads to aberrant inference (hallucinations, delusions).
Disconnection Hypothesis: frontal–temporal networks show reduced structural & functional connectivity.
Developmental views: neurodevelopmental trajectory; two-hit (genetic vulnerability + environmental insult e.g., cannabis).
Somatosensory & Motor Systems (Luigi)
Somatosensation
Tactile receptive field: skin area whose stimulation excites one neuron; size inversely ↔ tactile acuity (small on fingertip, large on back).
Somatotopy: adjacent skin ➜ adjacent cortical loci.
Cortical magnification: cortical area ∝ sensitivity/functional value, not body size (face > back).
Laterality & Integration
Two hemispheres connected by corpus callosum; early transfer between primary SI cortices enables bilateral coordination (hands neuro-functionally “close”).
Sensorimotor Interaction
Temporal & spatial proximity guide binding of tactile inputs with motor commands.
Motor Control Hierarchy
**Ventral (\"what\") vs Dorsal (\"how/where\") streams.
Motor equivalence: same goal via multiple muscle patterns.
Stages: goal selection → object identification → visuomotor transform → execution & online correction (e.g., mosquito trajectory).
Brain areas
Prefrontal (planning/executive), Premotor (movement preparation), Primary Motor (execution).
Motor Learning
Fast phase: rapid initial gains.
Slow phase: gradual optimisation & consolidation; duration task-dependent.
Interoception (Ruben)
Definition (classical): perception of signals from internal organs.
Expanded (functional/anatomical): includes any homoestatic or motivational signal conveyed via lamina-I spinothalamic pathway ➜ posterior insula (e.g., pain, temperature, affective touch).
Multidimensional construct
Afferent signal processing (brainstem reflexes).
Pre-conscious impact on cognition/behaviour.
Interoceptive accuracy (objective performance).
Interoceptive sensibility (subjective beliefs/attention).
Meta-awareness or insight (accuracy–sensibility correspondence).
Execution/attribution (causal explanation & action).
Measurement paradigms
Heartbeat Counting Task; Heartbeat Discrimination (synch/off-synch tones).
Respiratory load detection; adrenaline-induced arousal tracking.
Neural substrates
Brainstem, anterior cingulate, insula (posterior ➜ mid ➜ anterior gradient), secondary somatosensory cortex.
Skin as interoceptive organ
C-tactile (CT) afferents: hairy skin, optimally activated 1–10 cm/s stroking; project to insula; produce pleasant, socio-emotional touch.
Functional relevance
Emotion generation & regulation; conditioned fear without awareness (shock–image paradigm).
Self-awareness, decision making, motivational drive.
Visual Attention (Luis)
William James: selecting one train of thought, withdrawing from others.
Research questions: timing (early vs late), basis (features, space, objects), functional gain.
Empirical evidence
ERP (Luck et al.): cueing increases early components → enhancement of initial sensory processing.
Cocktail-party & Stroop: unattended info can intrude → evidence for late selection.
Perceptual-Load Theory (Lavie): processing depends on spare capacity; high load → early filter, low load → late.
fMRI Change-Detection (Beck et al.): successful detection recruits fusiform (object-specific), posterior parietal (spatial attention), DLPFC (executive).
Neural pathways
Dorsal (parietal): spatial attention & visual-motor control.
Ventral (temporal): object recognition (FFA for faces, PPA or adjacent fusiform for places).
Lesions: ventral damage → visual agnosia; dorsal damage → attentional/spatial deficits.
Conceptual models
Predictive Coding: brain as inference machine matching priors & evidence.
Object Files (Treisman & Kahneman): attention binds features into coherent objects.
Exam Logistics & Answer Sheet
Exam is multiple-choice (MCQ).
Fill in:
University / candidate number at top.
Put clear “X” inside chosen option box.
To change an answer: blacken incorrect box, mark new one.
No attendance code required for revision lecture.
Study Recommendations
Use lecture slides as primary framework; consult cited papers when concepts are unclear.
Master key terminology, classic paradigms, brain–behaviour links, and real-world examples.
Practise quick numerical / conceptual calculations (e.g., NN outputs, receptive-field logic).
Understand advantages & drawbacks of experimental methods (e.g., CVS vs GVS).
Keep biopsychosocial perspective when revising clinical topics.
Good luck!