● course setup
topic 1: brain basics
– chapter 1 (from modules to networks) + chapter 2
topic 2: neuroimaging methods
– chapter 3 + chapter 4 + chapter 5 (transcranial magnetic stimulation)
– chapter 3 section ‘mental chronometry in electrophysiology and cognitive psychology’ is optional but the
subsect ‘the spatial resolution of ERPs' is not
– chapter 4 section ‘analyzing data from functional imaging’ is optional
topic 3: vision
– chapter 7 + chapter 3 (distributed vs sparse coding)
topic 4: attention and action
– chapter 9 (the role of the frontoparietal network in attention) + chapter 10
topic 5: memory and language
– reading material on brightspace + chapter 12
– chapter 12 section ‘retrieving and producing spoken words’ is optional
● introduction
what can AI learn from neuroscience
– robustness, the brain demonstrates remarkable robustness, maintaining functionality even when parts
are damaged or lost; in contrast, AI systems often require all components to be operational to function
correctly
– robustness of inference, the brain is adept at handling variations such as rotation, noise, and unexpected
inputs (hallucinations); AI still struggles with these issues, often failing under adversarial attacks or when
presented with novel, untrained scenarios
– data efficiency, humans can learn new categories with just a few examples by leveraging prior
knowledge, unsupervised learning, and evolutionary constraints; AI typically requires vast amounts of
data to achieve similar learning
– energy efficiency, the human brain operates on approximately 15 watts of power, whereas a single GPU
can consume around 250 watts; this stark contrast highlights the brain's superior energy efficiency
– existence proof and inspiration, the brain is currently the only proven example of general intelligence,
understanding its workings can provide a blueprint for developing artificial general intelligence (AGI)
what can neuroscience learn from AI
– finding patterns in high-dimensional data, neuroscience benefits from AI's ability to analyse large,
complex datasets; machine learning techniques help identify patterns and insights that might be missed
, by traditional analysis methods
– using AI as a model for the brain, AI models, such as deep neural networks, can serve as computational
models for the brain; these models help researchers understand brain functions and test hypotheses
about neural processing
– task-performing computational models, AI provides task-performing computational models that mimic
brain functions, these models are used to understand how the brain performs specific tasks and
processes information
● topic 1: brain basics
the nervous system
integration, your body recieves sensory input and from there your nervous system processes that input, and
decides what should be done about it
motor output, the response that occurs when your nervous system activates certain parts of your body
the nervous system can be organized into several levels
– central nervous system (cns), your brain and spinal cord the main control system
– periphal nervous system, all the nerves that branch of from your brain and spine, they allow for
communication between the central nervous system en the rest of the body, it works in both directions
– sensory (afferent) division, picks up sensory stimuli
– motor (efferent) division, sends directions from your brain to muscles and glands
– autonomic nervous system, involuntary system that keeps the processes in your body going
– sympathetic division, mobilises the body into action (reflexes)
– parasympathetic division, relaxes the body
– somatic nervous system, voluntary system that rules your skeletal muscle movement
the nervous system is made up of nervous tissue
– glial cell types, provide support, nutrition, insulation, and help with signal transmission in the nervous
system, they make up of half of the brain mass
– in the central nervous system
– astrocytes, exchange of materials between neurons and capillaries
– microglial cells, immune defense against invading microorganisms
– ependymal cells, create, secrete and circulate cerebrospinal fluid
– oligodendrocytes, produce an insulating barrier called the myelin sheath
– in the peripheral nervous system
– satellite cells, surround and support neuron cell bodies
– schwann cells, produce an insulating barrier called the myelin sheath
– neurons, the cells responsible for receiving sensory input from the external world, for sending motor
commands to our muscles, and for transforming and relaying the electrical signals at every step in
, between
– cell body (soma), part of the neuron containing the nucleus and other organelles
– dendrites, branching structures that carry information from other neurons (receives signal)
– axon, a branching structure that carries information to other neurons and transmits an action
potential
types of neurons, how many processes extend out from the cell body
– multipolar neurons, one axon and multiple dendrites extend out from the cell body
– bipolar neurons, one axon and one dendrite extend out from opposite sides of the cell body
– unipolar neurons, only one process called the neurite extends from the cell body
neurons can be classified based on their function, so which way an impulse travels trough a neuron in relation
to the brain and spine
– sensory (afferent) neurons, transmit impulses from sensory receptors toward the central nervous system,
mostly unipolar
– motor (efferent) neurons, impulse moves from the cns to the rest of the body, mostly multipolar
– inter (association) neurons, impulse moves between sensory and motor neurons, mostly multipolar
action potential
action potential, a sudden change (depolarization and repolarization) in the electrical properties of the
neuron membrane in an axon
when a neuron is stimulated enough, it fires an electrical impulse down its axoms to its neighbouring neurons
– however, they only have on signal they can send, and it only transmits at one uniform strenth and speeds
– they can vary in frequency, the number of pulses
the body as a whole is electrically neutral, but some parts are more positive or negative than others, these
parts are seperated by membranes to build potential
– currents indicate the flow of positively or negatively charged ions across the resistance of your cells’
membranes
– voltage, the measure of potential energy generated by separated charges, the membrane potential
measured in mV
– current, the flow of electricity from one point to another, calculated by voltage/resistance
– resistance, whatever is getting in the way of the current
resting membrane potential (-70 mV), when resting, the neuron is more negative inside compared to the
space outside of it
– outside of the cell is made up of positively charged sodium ions (Na+)
– the inside of the cell holds positively charged potassium ions (K+), which are overpowered by bigger
negatively charged proteins
when a cell has a negative resting membrane potential, it is said to be polarized
this balance is made up thanks to the sodium-potassium pump in the membrane
, – for every two potassium ions that get pumped into the cell, it pumps out three sodium ions
– this created a difference is concentration of sodium and electrical charge, an electrochemical gradient
that has to be evened out
movement of ions is the key to all electrical events in neurons, so there are several ion-channels across
membrane
– voltage-gated channels, open and close in response to changes in membrane potential
– ligand-gated channels, open when a neurotransmitter latches onto its receptor
– mechanically-gated channels, open in response to physical streching of the membrane
graded potential, small openings from the ion-channels causes small changes in the potential, which doesn’t
do much
action potential, a big difference in the potential caused by stimuli, also called an impulse
– an action potential begins when a change in membrane potential caused by stimuli increases the
membrane voltage so that it reaches the threshold (axon hillock)
– action potentials are binary as they only fire or don’t fire, depending on whether the threshold
is reached or not
– once it hits the threshold, the voltage-gated sodium channels open
– the sodium ions travel towards the inside of the neuron, making the inside of the cell less negative
(depolarization)
– after some time the channels close (at around -50 mV), and the cell tries to restore balance and go back
to the resting membrane potential by opening the voltage-gated potassium (repolarization)
– during the repolarization, too many potassium ions leave the cell which leads to a membrane resting
potential of -75 mV (hyperpolarization)
– this imbalance gets restored and the neuron gets back to its membrane resting potential of -70 mV
– during the action potential, the neuron can’t respond to other stimuli (refractory period)
no matter the stimuli, the strength of the action potential is always the same, what does change is the
frequency of the action potential
– a weak stimulus tends to trigger less frequent action potentials
another thing that changes is the speed, the conduction velocity
synapses
the meeting point between two neurons, their strength and purpose lies in their connections
electrical synapses, nerve impulse is transmitted electrically via channel proteins, they are faster than chemical
synapses
– the connection between the sending neuron (pre-synaptic neuron) and the receiving neuron (post-
synaptic neuron) is called a gap junction
– this transmission happens very fast, but lacks control; if every synapse was electrical, the brain would get
overstimulated
