Task 2 Flashcards
What is creativity?
- multi-facet phenomenon
- manifests itself in different guises in different domains
- creativity in one domain might be different from creativity in another domain
What are theories of creativity?
- blind variation and selective retention
- defocused attention
- associative hierarchies and creativity
- primary process thinking and creativity
- creativity and arousal
What is the theory of blind variation and selective retention?
- there might be nothing special about creativity at all
- we think 16h a day -> the solution to a problem is a random thought (= emerges by chance)
neural-network:
thinking consists of the activation of nodes
- if the nodes are already strongly connected, thinking is routine and unsurprising
-the connection between nodes that are simultaneously activated are strengthened => creative insight
- weak/ indirect connections functions like hebbian learning
- genetic algorithms could be used to mimic insight
What is defocused attention ?
- differences in attentional capacity can explain individual differences in creativity
-> greater attentional capacities lead to higher creativity - creative ideas require consciousness
neural network:
consciousness can be divided into
- attention (= most activated nodes)
- short-term memory (= nodes that are activated less than in attention)
- nodes will be activated randomly after incubation
- more nodes can be simultaneously activated in creative people than in uncreative people
what is associative hierarchies and creativity?
- total activation in any layer of nodes is normalised or kept relatively constant in a network with recurrent lateral inhibition
- (stage 1 and 4 of creative process)
when attention is focused:
-> few nodes = highly activated
-> strong lateral inhibition on other nodes (= prevent them from becoming activated)
-> steep associative hierarchy = less creative - (stage 2 of creative process):
when attention is defocused:
-> activation = spread out among large number of nodes
-> less lateral inhibition (no node is extremely activated)
-> flat associative hierarchy = more creative
What is primary process thinking and creativity?
primary process thinking
- analogical
- autistic
- free-associative
-> the discovery of new combinations is more likely
- similar to large numbers of nodes being slightly activated
secondary process thinking
- abstract
- logical
- goal-oriented
- reality-oriented
-> most extreme form is deductive reasoning
= no creative insight possible, solution is part of premises
- best modeled as state of focused attention
What is creativity and arousal?
- each node receives information input from other nodes + non-specific input from the arousal system
- increasing arousal decreases creativity
- decreasing arousal leads to a shift from secondary processing to primary processing (= more nodes become activated)
- creative people show more extreme fluctuations in level of arousal
What is required to build a neural network of creative insights?
- we need to fill the nodes with a wide diversity of knowledge
- we need to present the network with a problem that it cannot solve
- we want the nodes representing the problem to remain partially activated (= so they can filter other nodes corresponding to perception and thought)
What is simulated annealing and creativity?
- intelligent random search, to find the best solution to a problem
- network aims at minimising energy
= energy is minimised for any given node when the constraints placed on it by other nodes are satisfied
= trial and error in the beginning - phase space of the network = all possible combinations of nodes being on or off
- for any scientific problem we are trying to find the global energy minimum
= an explanation that satisfies all the constraints imposed by the data and the rules of scientific methods - allows networks to avoid getting trapped in local minima
- optimising Parameter
- similar to gradient descent
example: 💎
high arousal
= low heat in chrystal
= likely to be at the global minimum)
the major theories of creativity and simulated annealing?
when the major theories of creativity are translated into neural-network theories -> they are identical
also seem to be identical with connectionist theories involving ismulated annealing
What is the neural network theory?
tries to explain how mental processes could be explained by neuronlike components
components:
1. a set of processing units/ neurons
2. a state of activation (activated nodes = consciousness)
3. a pattern of connections among nodes (EPSP/ IPSP)
4. input and output rules: how does a node add input? how does the output relate to the current activation?
5. learning rules (Hebbian learning)
6. Network environment
What is the common sense theory?
- the more knowledge about a certain domain, the more creative an indiviudal can be
- creativity is correlated with intelligence (= inverted U)
what are types of creativity?
- combinational creativity
= producing novel combinations of familiar ideas
= most difficult (-> machines miss cultural knowledge) - exploratory creativity
= adaptive in problem space
= move through space to discover previously undiscovered states of unexpectedly high value
= problem-space-search - transformational creativity
= actively transforms space
What is computational creativity ?
- emerging branch of AI studying the potential of computers to act as autonomous creators in their own right
- goal: for machines to exhibit human-level-creativity
- approaches:
= ignoring the need to define the phenomon objectively
= embracing the metaphorical foundations of creativity
= identifying an archetypal area of creative endavor
What is pastiche?
- creativity emerges through code that a human has written
= mimics rather than creates
= no innovation (the machine itself is not creative) - limitation to CC = humans are still needed
- pastiche based systems are not the goal of CC
