doing psych Flashcards
quantitative methods
‘traditional tools of science’
focus on development and testing explicit theories
qualitative methods
development of verbal theories
open-ended and explanatory (non - numerical)
quantitative theories
principle that explain a body of facts
specify the relations (often causal relations) between states
the goal is to understand / explain a system
must predict
what quantitative theories are not
description (they must specify structures of interest and pose and explanation for something)
a set of data (need an explanation of what data has been observed)
a diagram (unless it includes a description of logic underlying specified relationships)
david marr (1980) - science of mind - how does it work?
computational theory
representation and algorithm
physical implementation
what does computational theory consider? (marr, 1980)
what problems is it solving and why
what are the constraints on its solution
what is the nature of the problem that is getting solved/function being computed
what does representation and algorithm consider? (marr, 1980)
what information the system represents and how
what does it do with that information
what algorithm is it running on the information to get something useful from it
what is the input to the system, output and what stages does it go through in between
what does physical implementation consider? (marr, 1980)
how are these representations and algorithms realised in the hardware of the device itself (neurons of the brain)
what level of marr’s science of mind should we try to study cognition at?
implementation can be too abstract
computational links to biological capacity
rescorla wagner model
theory of learning via association
what kind of model is the rescorla wagner model
mathematical model
what are the 2 types of models in psych
mathematical and process models
explain mathematical models
models which use an equation to test a statement which associates an input (predictor) with an output (outcome)
explain process models
these models attempt to move down marr’s levels of science of mind into the representation & algorithm stage, and sometimes, ideally, the implementation stage
most common in cognitive and behavioural neurosciences
2 broad classes of process model s
symbolic process models
represent knowledge as symbolic data structures
there are sets of representations in our brains that can compose and make more complex representations
considered analogous to the computer metaphor of the brain
we have ordered instructions that change these representations
manipulate data structures with variablised rules
connectionist process models
represent knowledge as nodes in a network (nodes are analogous to neurons in brain)
knowledge is set in a node or collection of nodes
processing is carried out by passing activation between the nodes over weight of connection
symbolic models : data structures
basic or atomic elements that follow composition rules to make more complex structures
like a language and grammar rules
has variables (propoisiton) and operators and rules for putting these together
any 2 proposition can be combined by an operator
symbolic models : processes
symbolic operation on data structures - applying rules
if x is larger than y, then x can occlude y
the input doesn’t matter, so the rules will provide an answer
generally : cognition is like a traditional computer programme
how is cognition like a computer
mental representation = data structures
procedures = function or methods that operate on data structures
like R!
what is the prototypical symbolic model
a production system with 3 components that look like a computer programme
- data base (known facts)
- inference rules (ways of coming to a conclusion about something)
- executive control structure (how knowledge and rules interact) which rules are equipped and when - requires a specific algorithm)
states of production systems
current state
- current contents of database (known facts about the systems)
state space
- set of all possible states
goal state
- the state that you want the database to be in
state transition
- moving from one state to another (rules)
search
- the algorithm for traversing the state space and finding the best path for moving from current state to goal state ( deciding which rules to fire in which order)
symbolic models advantages
computational power
can define variablised and universally quantified rules (like monopoly)
symbolic models disadvantages
sometimes the rules are too rigid to capture human behaviour
fail to capture shades of meaning (cats are different to dogs, but are more like dogs than chairs)
not very automatic processing
do not answer the question of how we learn these rules
all learning occurs by application of the rules (which we have to assume are already built)
no graceful degradation with damage (system cannot reason with anything at all if 1 thing is gone, unlike brains)
no obvious neural implementation
connectionist models
composed of networks of interconnected nodes
nodes = simples processors which mimic neurons/populations of neurons
connections = the weight between the nodes
representation = the patterns of activation on the nodes/neurons
what do positive weights in connectionist processing mean
excitatory connections
connectionist model advantages
flexible processing (parallel constraints - not all or nothing, captures shades of meaning)
flexible representation -
- distributed representations capture semantic content
- permit automatic generalisation the network can do stuff there are no rules for (though, ability is limited)
graceful degradation with damage (take away nodes the network will still function - similar to our brains)
transparent neural implementation (easy to see how our brains make connections between information and meaning)
connectionist model disadvantages
not symbolic
ability to generalise is based on similarities (what you know about something can only generalise to objects like the ones you already know but this is not true of humans)
cannot represent or use variablised rules
- must learn about different instances
children generalise to utterly dissimilar examples sometimes
vulnerability definitions
DOH, 2000 = unable to take care of self
Smith, 2007 = those with enhanced risk of suicide, self-harm or harming others
how does the safeguarding vulnerable groups act 2006 define vulnerability
variety of criteria has been used in the construction of groups defined as vulnerable based on ways in which the groups are =
* marginalised
* socially excluded
* limited opportunities
* suffer abuse (psych, sex, finance, physical)
* suffer hardship
* suffer prejudice
* suffer discrimination
what kinds of things can lead to social exclusion
unemployment
lack of skills
low income
poor housing
high crime
bad health
family breakdowns
how can social exclusion occur
when problems are linked and mutually reinforced with each other and clustered in particular areas/neighbourhoods
- this creates a vicious cycle
