{[ promptMessage ]}

Bookmark it

{[ promptMessage ]}

11. Shultz pp. 111-172

11. Shultz pp. 111-172 - Page 111-172 Perceptual effects...

Info iconThis preview shows pages 1–3. Sign up to view the full content.

View Full Document Right Arrow Icon
Page 111-172 Perceptual effects Torque difference effect o Balance problem: equal number of weights on each side of fulcrum o Weight problems: distance constant, but one side has more weight than other o Distance problem: number of weights equal on both sides, distance varies o Conflict problem: one side has more weight but other has more distance Conflict-weight: Sides with greater weight go down Conflict-distance: side with more distance goes down Conflict-balance : scale is balanced o Rule-assessment methods show children progress through 4 different rule stages on this task o Children in stage 1 predict outcomes on basis of which side has more weight o In stage 2, use weight info, but begin to use distance info when both sides equal in weight o In stage 3, use weight ad distance, but become confused when weight and distance info conflict (so they guess) o In stage 4, good at wide range of problems (meaning they may be comparing torques) o Torque-difference effect: balance scale problems where there are large torque differences are easier for children to solve than those with small torque differences This perceptual effect (because it depends on how the balance scale looks) cannot be explained by production rules Torque-difference effect is found in children in each of the stages and not reflected in symbolic rules at any stage But cc networks naturally capture this where networks performed better on problems with large torque difference Some BP networks can also Problem size effect in conservation
Background image of page 1

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full Document Right Arrow Icon
o This is also unexplained by rules o Children develop conservation with small numbers before they do with large numbers o Rules cannot account for this but cc networks trained on conservation problems showed it at intermediate levels of training (not early when performance was poor on all problem sized or late when performance good on all problem sizes) Understanding neural representations Drawback of neural network modeling is that the knowledge representations developed by a network are quite opaque. We may know that the knowledge representations in there does a good job of covering psychological data and yet we may be ignorant of what the knowledge representation is like But we need not be ignorant for long. We can try to extract the rules used by the networks as they process info and generate responses. Neural networks are rule like even if they are not rule based. It has only unit activations and weights and can come to behave as though it has rules. Chapter 4 – Developmental Transitions Proposed transition methods Piaget’s theory of transition o Can map Piaget’s theory to computational features of cc Assimilation (assimilate info from environment by distorting it to fit current cognitive system) – forward propagation of inputs Accommodation (adapt cognition to external environment) – output weight adjustment Equilibration (maintain balance between accommodation and assimilation) –
Background image of page 2
Image of page 3
This is the end of the preview. Sign up to access the rest of the document.

{[ snackBarMessage ]}

Page1 / 10

11. Shultz pp. 111-172 - Page 111-172 Perceptual effects...

This preview shows document pages 1 - 3. Sign up to view the full document.

View Full Document Right Arrow Icon bookmark
Ask a homework question - tutors are online