Similarity

Question 1

Why is similarity important?

Answer 1

• categorisation (e.g., colour of the creature deciding the category that its in)
• learning (e.g., learning whether previous knowledge can be implied into the new/existing context) (The Duncker’s radiation problem)
• gestalt perception (likelihood of grouping similar items into a group)
• memory
  (likelihood of remembering depends on similar items that are presented)
  (Even if the word dream didn’t appear, we think it did due to the similar semantic relationship with other words.)
• generalising (generalising new items based on similar properties of previous items)
• witness identification

Question 2

1. what is similarity
2. On what basis to evaluate similarity
3. how to measure similarity

Answer 2

1. similarity = the sameness between two objects.
   sameness = the two objects evoke the same action/consequences
2. perception
   goal-based
   conceptual
   relational
   analogies
3. scale
   response time
   confusability
   AXB force choice
   stimulus arrangement

Question 3

What are the methods that are examined in the Hout, Goldinger & Ferguson Experiment

Answer 3

scale
AXB Force Choice
Stimulus arrangement

Question 4

Explain how the multidimensional scale is used in the experiment and the results

Answer 4

The further the objects (i.e., the clock and the bugs) are apart from each other on the scale, the more different they are perceived by the participants

Then, the deviance between the participants’ scale and the ideal scale (ideal stimulus space) is measured
> the higher the deviance, the more inaccurate the method is

Turns out the stimulus arrangement method = lowest deviance = the most accurate

Question 5

What does the Hout, Goldinger and Ferguson Experiment measure

Answer 5

the accuracy of the three methods measuring similarity

Question 6

When will the stimulus arrangement not work? and what is the alternative

Answer 6

when there are too many stimulus for arrangement
can have the participants to do a portion of the stimulus
then combine the data of all participants
if the number of participants are enough, should have enough data for all stimulus
then plot onto the MDS to understand the underlying relationship between the stimulus

Question 7

how many dimensions do people consider when viewing the rocks and other real world objects respectively

Answer 7

8-10, 47-48

Question 8

Explain geometric model of similarity (and two additional characteristics)

Answer 8

human mind subconsciously plot items on a graph according to their degree of similarity, even though the stimulus are given separately
the closer the items are, the more similar they are
also shows the dimensions that human took into account when distinguishing similarity
e.g.,
faces > adiposity and age
boxes > height and bar position (height makes more difference than bar position)
colour > different from physical representation (instead in a circle)
musical notes > different notes on a same octave (chroma) and same note but on different octave (latter = more similar)
emotions > hedonic valence (goodness/badness) and the potency (strength)

Question 9

flaws of the geometric model

Answer 9

symmetry constraints
mathematically
A > B = B > A
however, this might not be the case
the surgeon is like the butcher /= the butcher is like a surgeon

triangle inequality
mathematically
d(A,B) + d(B,C) > d(A,C)
however, this might not always be the case
panda = koalas
kangeroos = koalas
but panda /= kangeroos

can be solved by limiting the attention to only one factor

Question 10

explain the feature models

Answer 10

humans view similarity based on the number of similar features that the objects show
more similar features = more similar

common element model
only focuses on the similar features

tversky’s model
also focuses on different elements
selective attention comes into play
if the person decides to put more attention to the differences = less similar
if the person decides to put more attention to the similarities = more similar

solved symmetrical constraints and asymmetrical similarities

Question 11

explain structural alignments

Answer 11

based on the placements of the objects in the pictures
more MIPs = more similar
MOPs also increase similarities, but to a smaller extent

transformation
more transformation = less similar

Question 12

What are the two issues that similarities fail to explain in decision making

Answer 12

the degree of variability of an object and dissociation between categorisation and similarity

Question 13

whose theory is the closest with ‘relational processing’?

Answer 13

Goldstone and structural alignment