L15 - Prototypes - Exemplars and Category Learning Flashcards
Multi-Dimensional Scaling (MDS) transfers similarity into ______
distances
What is the major assumption of the Prototype view?
Our conceptual knowledge is stored in the form of an abstraction
In the prototype view, how do we represent a category?
We abstract out the central tendency of a category on the basis of all our experiences with the category
- Category representation consists of a summary of all the examples of the category called the prototype*
- this is an abstraction*
What does the exemplar view suggest about category representation?
That we simply store in memory every example of a given category that we encounter.
According to the exemplar view, our conceptual representation consists of -
all the individual members of a category, known as exemplars.
How can you view the prototype and exemplar views?
As a continuum
Prototype = total abstraction
Exemplar = zero abstraction
What is the positives and negatives of the prototype view?
Positive - it reduces memory load
Negative - memory load reduction comes at the cost of information
What is the positives and negatives of the exemplar view?
Positive - useful because it retains specific information
Negative - but it comes at the cost of increased memory load
What is the problem with the prototype view?
Can we store everything we know about a category member as a ‘prototype’?
- What does it look like, is it merely a set of features?
What is the problem with the exemplar view?
Does it seem likely that store in my memory every example of a category that I come across?
Each view seems implausible at its extreme
What is the issue with the fact that both exemplar and prototype models at their extreme are implausible?
That we need to choose between one or the other in order to implement a computational model
The family resemblance model is an prototype or exemplar model?
Exemplar
The family resemblance model predicts typicality as a function of featural overlap between category members (exemplar)
The poymorphous concept model is an prototype or exemplar model?
Prototype
The polymorphous concept model predicts typicality as a function of featural overlap between category members and an abstracted feature list representing the category name (prototype)
Typicality can be predicted as a function of the distance between
each category member in a multidimensional space and the central
tendency of that category
Is this using a prototype or exemplar model and why
Prototype
Because it is a measure of central tendency
Typicality can also be predicted as a function of the mean distance between each category member in a multidimensional space and each other category member in that space
Is this using a prototype or exemplar model and why
Exemplar
Involves all the members of the category
What type of stimuli do category learning experiments use?
Tend to employ stimuli that bear little resemblance to ‘real-world’ categories, but can be easily manipulated
– Hue, Brightness, Saturation, Size, Shape, etc
What are the two phases of category learning experiments?
1. Learning phase
2. Transfer phase
In category learning experiments, what does the learning phase consist of?
4 steps
In category learning experiments, what does the transfer phase consist of?
After the learning phase, you introduce new stimuli to the participants and ask them to categorize them
Record how they categorise the stimuli based on on what they experienced in the learning phase
What happens when participants have to attend to “multiple category dimensions” in category learning experiments?
They find it harder to learn the categories and make more errors
The less dimensions to learn the quicker we learn the categories
How do category learning experiments give us insight to?
This gives us insight into our ability to generalize from a stored category representation to novel stimuli.
• Manipulating the category structure allows us to test different theories about the processes underlying categorization and generalization.
Many category learning experiments consist of a third stage, what is this stage?
Similarity Rating Stage
The empirical categorization decisions and/or response latencies are computationally modelled using similarity rating in order to generate MDS scaling representation of the stimulus space.
How do we computationally model category learning experiments?
Involves using a similarity rating in order to generate a Multi-dimensional scaling representation of the stimulus space.
What is the most well known and commonly used model for the third stage of category learning
Generalized Context Model (GCM) (Nosofsky, 1986).
What does Generalized Context Model (GCM) (Nosofsky, 1986) state about the probability of a stimulus being categorised as a member of a category.
Is this model prototype or exemplar?
The probability of a stimulus being categorized as a member of a given category is a weighted function of the distance between the target stimulus and the members of the two categories in the space
(exemplar)
We use distances in MDS to convert into similarities
Describe The MDS-based Prototype Model (MPM) for calculating the probability of a stimulus being categorized as a member of a given category.
(Minda & Smith, 2001)
The probability of a stimulus being categorized as a member of a given category is a weighted function of the distance between the target stimulus and the prototypes (central tendencies) of the two categories in the space
Is the GCM (exemplar) or the MPM (prototype) better at predicting which member will belong to a category
Both do well - however GCM does better than MPM
Exemplar models seem to do slightly better at representing how we apply membership to categories.
Does this mean that this model is right? Why?
No
The true repesentation likely lies in between no abstraction (exemplar) and full abstraction (prototype)
The true way we representation lies in between exemplar and prototype - what model was developed to account for this?
Vanpaemal & Storms (2008): Varying Abstraction Model (VAM)
How did Smits, Storms Rosseel and De Boek (2002) test whether GCM or MPM models work when we categorise things in real life?
- Data set: pictures of 79 well-known fruits and vegetables, and 30 novel stimuli (mainly tropical fruit and veg)
- One group of participants made pairwise similarity ratings
- One group asked to categorize all 109 stimuli as either fruits or vegetables
What were the results of the Smits, Storms Rosseel and De Boek (2002) test which tests whether the GCM matched how people categorized objects in the real world?
The GCM is accurate for accounting how human categorize.
Similarity can account for these types of decisions
MPM also does equally well
