· Slightly better spatial resolution, especially better for deeper structure because its not relying on scalp sensors; much better for deeper, subcortical strucutres

o Magnetic resosnance imaging (MRI): a patient is placed in a large, powerful magnet that produces a strong magnetic field that influences how atoms spin

§ A radiofrequency current pulsed through the patient causes the atoms to spin out of equilibrium

§ When the pulse is turned off, MRI sensors detect the energy released as atoms realign with the magnetic field

§ MRI tells us a bout water rich (i.e. soft) tissues

§ Get a snapshot of the brain from a living person à structural information

§ Can reconstruct 3D images

§

§ Pro/con

· MEG is activity; vs MRI is structural information

· Costly (compared to like an xray)

· Pro: doesn’t use radiation

· Better pictures of soft tissue vs x-ray

· Very uncomfortable: can’t move, claustrophobic, very loud; makes it hard to implement for many populations

· Because its loud, its hard to present auditory stimulus so it can’t be used for audition

o Function MRI (fMRI): magnetic pulses pick up evidence of demand for more oxygen in the brain, creating a blood oxygen level-dependent (BOLD) signal

§ More active areas need more blood (oxygen)

§ Can record the activity of the living brain à functional information

§ Stimulus evoked activity minus baseline = change caused by stimulus (i.e. substractive)

§ Pro/cons

· Low temporal resolution; because recording blood flow; neurosn have to use up energy, then vascular system needs to supply more blood; so ther eis adelay

· Indirect measure; bloodflow response to neuron activity

· Very helpful for subcortical structures

· noninvasive

§

o Positron emission tomography (PET): a small amount of tracer (a biologically active, radioactive material) is injected into the patient’s bloodstream (2-deoxyglucose, 2DG)

§ Specialized camera detects the radiation emitted from brain regions using more of the tracer (i.e. metabolically active areas)

§ E.g. type of glucose that the brain can use à where is it directed during various tasks

§

§ Pro/con

· Poor spatial resolution

· Can use auditory stimulus

· Can look at deep structures

Modeling as a Method

· Mathematical models use mathematical language, concepts and equations to closely mimic psychology and neuronal processes with mathematical precision

o Example: H&Hs model described how action potentials in neurons are initiated and propagated

o

· Computation models use mathematical language and equations to describe steps in physiological and/or neural processes (often implemented on a computer)

·

o The real world is more structure, redundant and predictable vs in a field of random noise, knowing one spot tells you nothing about its neighbour

· Computational models

o Efficient coding models: assume that sensory systems become tuned to predictability in natural environments in ways that economically encode predictable sensory inputs while highlighting inputs that are less predictable

§ Like how computers store and compress data

§ Compress resuntant information, keep the bits that you care about

§

o Bayesian models: employ Bayesian statistical models – which assume that earlier observations should bias expectations for future events – to build a model of the world (sensory inputs)

§ Models predict future events (predictive coding). If predictions don’t match inputs (prediction error), the model is adjusted to improve future predictions

§ What you experience in the world might help you understand/predict the future

o Artificial neural networks: comprised of layers of heavily interconnected computational units analogous to neurons massively connected with one anther through their axons, dendrites and synapses; Strength of connections can increase or decrease with experience akin to learning

§ Includes AI, machine learning, neural networks, deep learning

§

§ Deep neural networks: have many ayers of units (nodes) with millions of connections; very good at taking lots of info and classifying it into categories

· This is the AI technology responsible for google home or fcial recognition software

§ A neural network has

· Inputs

· Weights: how important is this input to the outcome?

· Threshold: minimum output of a single node in order for data to be sent to the next layer

· Output

§ Deep neural networks are feed-forward and some can also be trained through feedback

§

·

Thresholds and the Dawn of Psychophysics

· Classical psychophysics

o Pioneered by german physicist and philosopher Gustav Fechner

o Considered the true father of experimental psychology

o Pioneered psychophysics: the study of quantitative relationships between physical stimuli and psychological experiences

· How can we describe the relationship between mind and matter?

o Why relate physical stimuli to perceptual experience using emthematical mdoels?

§ If we can quantify what the standard is, we can identify when people may be experiencing deviations (e..g usually hear X sound, or see at X distance)

o Function: mathematical description of how one variable is related to another; generally expressed as a formula

o

o Why not start at origin?

§ If theres no stimulus, theres nothing to detect

§ The gap represents the threshold; minimum value of stimulus before it is detected

· Classical psychophysics is centred on the idea of thresholds

o Absolute threshold: minimum stimulus level required to be registeres by the brain as a sensory event; where the function begins

§ Subthreshold: below the level of detection

§ Suprathreshold: above the level of detection

§ Examples:

·

· How can we measure thresholds?