Please enable JavaScript.
Coggle requires JavaScript to display documents.
Creativity and Connectionism (Martindale (Neural Network Theory Components…
Creativity and Connectionism
Martindale
Neural Network Theory Components
Processing units or nodes that are similar to neurons but not as complicated
a state of activation beyond a threshold- the one or two most activated nodes correspond to whatever is the focus of attention & less activated nodes content of STM
Connections among nodes can be excitatory or inhibtory > constitute LTM
Input and output rules concerning how a node adds up ins input and output relate to current activation
Learning Rules via Hebbian Learning
environment – each module in a network is organized in several layers with vertical connections constituting excitatory connections & lateral inhibition operating on each layer
cognition is parallel, all nodes do something at the same time- in conventional computer can only do one thing at a time
creative ideas must be novel und useful> always new combations of old ideas
4 Stages of Creative Idea:
Preparation
– Work intensively on a problem without arriving at a conclusion (manipulate ideas)
Incubation
– Set problem aside as no progress happens
Illumination
– Solution to the problem ‘pops up’ (ideas not previously thought to be relevant)
Verification
– Scrutinize the idea
Common Sense and some Facts
Having knowledge of a wide variety of things/ wide range of interests is a good predisposition to creativity: it allows combining ideas from different disciplines
Hopfield, a physicist, came up with neural network theory by comparing spin glasses (physics) with neurons in the brain (cognitive science) and applied physics equations to the brain
BLIND VARIATION
VS SELECTIVE ATTENTION
neural network analogy: thinking= activation of nodes > Two strongly connected nodes= routine thinking
If two weakly/ indirectly connected nodes are active at same time, the arousal system activates the cortex with non- specific activation, leading to an increase in activation of the nodes strengthening their connection (Selective retention)
Creative Insight
rezipier for insight
Fill node with diversity of knowledge
Present problem the network cannot solve
Keep nodes representing the problem partially activated so they can filter other nodes
If other nodes give a hint at the solution, connections are strengthened > solution pops up! >Be in low- arousal state to have many nodes activated
As to not have a problem space that is too large one must cut down the ‘associative horizon’ (Eysenck) and vary between low- & higher- arousal states
Neural Network analogy
: Consciousness= Attention (most activated nodes) + STM (less activated nodes)
Creative people may have more nodes that are simultaneously activated> Prep stage: Too focused attention (few nodes highly activated), incubation: in creative people nodes coding the problem remain partially activated while in uncreative people they become deactivated, Illumination: A node activated relates to the (partially activated) node from the problem fully reactivating it, verification: focused attention
Defocused Attention
: Differences in attentional capacity explain individual differences in creativity: The greater a :ttentional capacity, the more likely combinational leap that is needed for creativity (the more ideas one can attend to simultaneously, the more likely their combination)
Creativity requires less focused attention
ASSOSIATIVE HIERACHIES
During Focused attention, few nodes are highly active and exert strong lateral inhibition, preventing other nodes from being activated
STEEP HIERACHY- Less creative
Secondary Process – abstract, logical (deductive), goal- oriented >found during creative elaboration/ verification
a.= best modeled as a state of focused attention, with few nodes being strongly active
Assuming that the total amount of activation in a layer of nodes is kept constant, then one node has all activation at a high level of arousal (steep hierarchy/ focused attention/ secondary- process)
when attention is focused, the wird will strongly activae only a view nodes and the assosoative hiarachy goes up
During Defocused attention, there is less lateral inhibition, thus nodes become more activated (STM nodes)
FLAT HIERACHY- More creative
Primary Process – thinking in analogical, autistic & free association > found during creative inspiration
a.= Large numbers of nodes are weakly & equally activated (little lateral inhibition)
when attention less focused, more nodes active but to lesser degree and assosiative hierachy flat
creative people can more alternate between 1. and secondary process cognition
CREATIVITY AND AROUSAL
the more you are aroused the more behavior is stereotypical (non- creative)
Increases in arousal make behavior more stereotyped & decreases in arousal make behavior more variable (Hull)
low arousal= more nodes active=thinking from secondary to primary
In neural network model, activation of a node is determined by: Input from other nodes + nonspecific input from the arousal system
STUMULATED ANEALING
Hopfield showed that such a network evolves to minimize energy, reaching a state of fixed- point attractors
GLOBAL MINIMUM
: achieved for any given node by satisfying all constraints placed on it by other nodes (two positively connected nodes have minimal energy if both are on)
With a large number of nodes, the network has to try various combinations before findings the global energy minimum and iteratively update the activation of the nodes until the state has been reached
Problem: some constraint (connections or interconnections)do not apply to brain: caught in local minimal
Solution: Stimulated annealing:network anneals periodically (oscillate between high/ low temperature)
Starting with the temperature very high & nodes go on and off in a random fashion and gradually lower the temperature
At high temperatures – nodes can increase & decrease contribution to total energy which allows the system to crawl out of local minima> primary process fashion
the analogy between simulated annealing & the explanation of creativity->temperature is the inverse of cortical arousal: High temperature= Low arousal
Comparing Hopfiled and new theory: Low temperature= close to a step function, with node being on if input is above 0 and off it input is below 0
Higher temperature= Nodes behave more randomly (node can take value of -1 if input is positive of +1 if input is negative)
Brain Network and creative cognition: dynamic interaction of default + control network> default influences generation of candidate ideas, but control network can constrain+ direct process to meet task-specific goals via top-down monitoring+ exectuvie control
KLONDIKE SPACE