Foundations in Psychology - Semester 1, Lecture 1

WHAT IS PSYCHOBIOLOGY?

• Study of the brain, human or animal behaviour & the relationship between the two
• Application of principles of biology to the study of mental processes & behaviour
• Goal - Understand the biological processes underlying psychological phenomena
• Research - How psychological factors like cognition, mood & appraisal combine with biological events like stress physiology, changes in brain function & pharmacological effects to shape human experience
• *See flow chart

WHAT IS COGNITIVE PSYCHOLOGY?

• Study of mental processes
• Key Features:
• Main approach to experimental psychology​​​​​​
• - In cognitive psychology which investigates memory, language, perception & problem solving
• - Also used for other areas like social & developmental
• Emphasises active mental processes
• - The brain is seen as an information processor using the analogy of mind to computer
• - Mental processes are based on discrete modules
• Uses experimental methods as well as computer modelling & neuropsychology

WHAT IS COGNITIVE PSYCHOLOGY?

• Working Memory
• - Central executive: controls/directs attention
• - Visuo-spatial sketchpad, phonological loop & episodic buffer
• Temporal Perception
• - Internal clock/pacemaker
• Schemas
• - Mental representation of thought/behaviour organisation
• - Activated in relevant situations
• *See flow charts

COGNITIVE PSYCH VS PSYCHOBIOLOGY?

• Cognitive Psychology: How does the mind do it?
• Psychobiology: How does the brain do it?
• Both can use neurons to describe mind
• Mental processes vs physical biology

EARLY BEGINNINGS

• Early Philosophers:
• Plato (387 BC) suggested the brain was the seat of all mental processes
• Galen (AD 130-200) proposed a theory of brain function based on ventricles
• Plotinus (AD 205-270) said that the soul is separate from the body (dualistic understanding of mind & body)

17TH-19TH CENTURY

• Descartes (1596-1650)
• - Dualism (& the mind/body problem): interested in how does physical matter
• - Mechanistic view of the body (eg. reflexes)
• - Mind does not follow the laws of nature
• - Mind & body can influence each other
• - Pineal gland as the interface between body & soul
• Debates during the 18th & early 19th centuries
• - Those who believed that brain function could be localised to particular brain regions
• - Those who believed the brain acted as a whole

DEBATE - EARLY-MID 19TH CENTURY

• Franz Joseph Gall (1757-1828) & Johann Spurzheim (1776-1832) developed phrenology (the idea that behaviours & characteristics could be deduced by the pattern & size of bumps on the skull
• ​​​​​​​
• Marie-Jean-Pierre Flourens (1794-1867) believed that parts of the brain had separate functions but each of these areas functioned globally as a whole

DEBATE - LATE 1800-EARLY 1900

• Broca (1824-1888) & Wernicke (1848-1904) in the late 1800s provided strong data to support localisation of function, they identified specific areas of the brain central in the production & comprehension of speech
• Golgi (1873) discovered silver staining to identify neuron structure

WILHELM WUNDT

• Father of experimental psychology
• Published 'Principles of Physiological Psychology' (1873)
• Consciousness - analysis of the subjective experience of the mind
• ​​​​​​​
• Opened first psychology laboratory in Leipzig, Germany (1879)
• ​​​​​​​
• Scientific techniques to isolate subtle processes

EBBINGHAUS & MEMORY

• Systematic & controlled study of memory in laboratory
• Devised methods to measure memory & the speed with which forgetting occurred
• ​​​​​​​​​​​​​​​​​​​​​
• Forgetting & learning curves

EARLY 20TH CENTURY ADVANCEMENTS

• Growing evidence for localisation in the early 20th century, direct evidence from motor & sensory maps in the brain from early work of Wilder Penfield
• Stimulating the temporal lobes could elicit meaningful integrated responses
• ​​​​​​​​​​​​​​
• Physical basis of memory, an 'engram'

BIOLOGY OF MEMORY

• Karl Lashley (1950)
• - Searched for the engram, the physical location of a memory
• - Trained rats to solve maze then cut out pieces of their cortex & re-tested their memory of the maze
• - Partial memory retained

WHAT ABOUT COGNITION?

• Behaviourism dominant until the late 1950s, B. F. Skinner
• Scientific approach
• ​​​​​​​
• All our actions are the consequence of the response, reinforcement relationship
• ​​​​​​​​​​​​​​
• Including 'Verbal Behaviour' (1957)
• ​​​​​​​
• Chomsky's (1959) critique of 'Verbal Behaviour' is that it doesn't account for child sentence creation & an innate ability must exist

WHAT ABOUT COGNITION?

• Bartlett (1886-1969), Piaget (1896-1980) & Lewin (1890-1947), early pioneers
• ​​​​​​​
• WII shifted the need to look at cognitive skills
• ​​​​​​​
• 1950s technological improvements gave a theoretical (information processing) as well as methodological boost
• ​​​​​​​
• Studies software but not hardware of the brain

COGNITIVE NEUROSCIENCE

• Psychobiology + Cognition = Cognitive Neuroscience
• Study of the neurological basis of cognitive processing
• Concerned with the scientific study of biological substrates underlying cognition with a specific focus on the neural substrates of mental processes
• It addresses the questions of how psychological/cognitive functions are produced by neural circuits in the brain
• Computational neuroscience - detailed simulation of neuronal mechanisms

WHY STUDY THEM?

• It's important to know how the brain works & how it affects our behaviours
• ​​​​​​​
• Implications for many areas of psychology, abnormal psychology, developmental psychology, forensic & investigative psychology, HCI, etc.

METHODOLOGIES

• EEG's (electroencephalogram's) & ERP's (event related potentials)
• TMS (transcranial magnetic simulator) - a magnetic pulse applied to a brain region to 'switch it off'
• ​​​​​​​
• CAT (computerised axial tomography) - series of multiple x-rays
• ​​​​​​​​​​​​​​
• PET (positron emission tomography) - radioactive tracer introduced to body & traced
• ​​​​​​​
• fMRI (functional magnetic resonance imaging) - measurement of blood flow through brain regions

COGNITIVE PSYCHOLOGY

• Methods of Investigation
• Experimental methods - lab studies
• Simulations
• ​​​​​​​
• Case studies on acquired & developmental deficits

LANGUAGE & CONNECTIONIST MODELS

• Overview
• The acquisition of the past-tense of verbs in English - over-regularisations & the U-shaped profile of learning
• Computational model 1 - the symbolic model of the past tense learning
• ​​​​​​​
• Brief introduction to connectionist or neural network models
• ​​​​​​​
• Computational model 2 - a connectionist neural network model of the past tense learning

PAST-TENSE IN ENGLISH

• To form the past tense of regular verbs, -ed is added
• The past tense of irregular verbs is not formed by adding -ed
• When children add the -ed at the end of an irregular word, they are producing 'over-regularisations'
• ​​​​​​​​​​​​​​
• Some authors claim that these over-regularisations provide evidence that children are learning rules

IRREGULAR VERBS - KUCZAJ (1977)

• There are about 150-180 irregular verbs that can be classified into 5 different categories
• - Internal vowel change
• - Internal vowel change & addition of a final dental consonant
• - The final consonant is changed to a dental consonant
• - No change
• - Total change

U-SHAPED PROFILE OF LEARNING

• At an early age, children produce correct forms of past tense for both regular & irregular verbs
• When children are about 3/4 years old, they start producing over-regularisation errors of the type 'go-ed' & 'went-ed'
• ​​​​​​​
• At a later age, children produce correct forms of past tense for both regular & irregular verbs
• ​​​​​​​
• This is called the U-shaped profile of learning

COMPUTATIONAL MODEL 1

• Computational Model 1 = The Symbolic Model of the Past Tense Learning
• Symbolic Model: Dual-Route Model
• Two mechanisms
• - A memory storage device containing the past tense of irregular forms
• - A rule-based system that adds the '-ed' form to the stem of a regular verb to form the past tense
• *See flow charts

SUMMARY OF THE SYMBOLIC MODEL

• Two mechanism
• - A memory storage device containing the past tense of irregular forms
• - A rule-based system that adds the '-ed' form to the stem of a regular verb to form the past tense
• Explanation of development​​​​​​
• - At the beginning, all past tense verbs are stored in the memory storage device
• - The beginning of over-generalisation is explained by the interference of the two mechanisms
• - With time & practice, the two mechanisms discover the correct division of verbs into regulars & irregulars

THE SYMBOLIC MODEL & POSITIVES

• Evidence
• Normally developing children make mistakes on irregular forms only
• Frequent irregular verbs are also less likely to be over-regularised
• Specific Language Impairment vs Williams Syndrome (Pinker, 1994)
• - Individuals with SLI make mistakes on the irregular verb by over-regularising them
• - Individuals with Williams Syndrome have only a memory storage that stores both the regular & irregular verbs
• This dual architecture is innate
• It can account for the U-shaped profile of development
• Takes into account why some irregular verbs are more likely to be over-regularised

POSITIVES & LIMITATIONS

• Limitations
• The model does not tell us how children learn the '-ed' rule
• Kuczaj (1977) indicated that irregular verbs could be classified into 5 different categories depending on the type of rules that lead to the formation of the irregular past tense
• Cross-linguistic variations
• - In Arabic, the regular form of plural constitutes less than 20% of the plurals in the language
• - Norwegian has 2 ways of forming the past tense in regular verbs (add '-te'/'-ede'), plus 100 irregulars
• - Both mean it is very difficult for children

CONNECTIONIST MODELS

• Also called neural network models
• Altman (1997) compares a neuron to the structure of a sunflower

NEURONS & THEIR ACTIVATION

• Main Characteristics
• Characteristic 1
• Neurons send impulses to the other neurons to which they are connected
• The rate at which impulses are sent corresponds to the 'strength' of the signal
• Characteristic 2
• The impulse from one neuron can either
• make it more likely that another neuron will become active & pass on the activation (excitatory) or less likely that another neuron will become active (inhibitory)
• Characteristic 3
• The connections from one neuron to other neurons can change in response to the surrounding activity

COMMUNICATION BETWEEN NEURONS

• Excitatory Postsynaptic Potential (EPSP) = may trigger a new action potential by depolarising the neuron
• Inhibitory Postsynaptic Potential (IPSP) = may inhibit a new action potential by hyper-polarising the neuron
• ​​​​​​​
• Summation (things added together) of the postsynaptic potentials determine whether a postsynaptic will fire

BASIC PRINCIPLES OF THIS MODEL

• Connectionist/neural network models are computer models that try to simulate how the neurons in the brain communicate & learn
• These computer neurons are much simpler than real ones
• The computer allocates a number to a neuron to indicate how active each neuron is, this is similar to the activation of a real neuron
• An algorithm calculates how active a neuron is, which will depend on how many neurons connected to it are activated & the strength of that activation
• The connection strength between two neurons is a number
• If the number is positive, then it is an excitatory connection, however if the number is negative, then it is an inhibitory connection

MORE BASIC PRINCIPLES OF THIS MODEL

• These models offer an alternative to the rule-based accounts of the inflectional system
• ​​​​​​​
• The architecture of the connectionist models is not innate
• ​​​​​​​
• Using the same principles as the Behaviourist perspective, connectionist models learn from the environment

COMPUTATIONAL MODEL 2

• Connectionist/Neural Network Model of the Past Tense Learning
• Plunkett & Marchman (1993-1996)
• This is a network that tries to explain the past learning & processing of the past tense of verbs in English
• Architecture of the 'feed-forward network' with 20 input units, 30 hidden units & 20 output units

PLUNKETT & MARCHMAN (1993-1996)

• The aim of the network is to find the right set of connections between the neurons that will activate '-ed' when a regular verb is presented (excitatory) but will switch them off when an irregular verb is presented (inhibitory)
• ​​​​​​​
• During training, it is likely that there would be temporary patterns of interference between the right connection for the regular verbs & the right ones for the irregular verbs, this leads to over-regularisation errors

PLUNKETT & MARCHMAN (1993-1996)

• Learning
• Initially, the model learned the past tense of 10 regular & 10 irregular verb stems which reflects the current estimates of the balance between regular & irregular verbs in children's early vocabularies​​​​​​​
• After that, the vocabulary was increased to mimic gradual uptake of verbs by children​​​​​​​
• The network saw a total of 500 verbs, 90% of these being regular & verbs introduced early in training had a higher frequency​​​​​​​
• The output pattern is compared to a teacher signal which specifies the correct past tense form of the current verb

PLUNKETT & MARCHMAN (1993-1996)

• At the beginning, there are no errors
• The overall rate of over-regularisation errors is between 5% & 10%
• Over-regularisation recurs throughout the training period, they are not restricted to a particular stage of development
• Over-regularisation of high frequent irregular verbs is very rare
• The phonological properties of some irregular verbs seem to block over-regularisation
• A very small number of irregularisation errors are observed

PLUNKETT & MARCHMAN (1993-1996)

• Evidence
• The pattern of behaviour of the network is similar to the data found by Marcus et al (1992) when they studied the spontaneous speech of 83 English-speaking children
• The network demonstrates that we can simulate patterns of regular & irregular verbs without the need for an innate architecture

USES OF NEURAL NETWORK

• Over-regularisation errors are not observed until the vocabulary size reaches around 120 verbs
• ​​​​​​​
• While the network has been exposed to less than 50 verbs, it treats each novel verb stem unsystematically as the net still hasn't extracted a 'representation'
• When the vocabulary expands from 50 to 120 verbs, the performance of the network improves by around 70%, there seems to be a critical period where the network moves from a 'rote representation' to a more systematic 'rule representation'
• ​​​​​​​
• The number of regular verbs during training is crucial for the success of the network