PY1101 Lecture Notes - Lecture 5: Eleanor Rosch, Soltyrei, Risk Aversion
14 Jun 2018
School
Department
Course
Professor
Conceptual Knowledge and Decision Making
Conceptual Knowledge
- knowledge enabling you to recognise objects, access semantic information about
them (i.e facts about them), and make inferences about their behaviour/functions
- contains many concepts – knowledge about specific types of objects
o concepts are hierarchically ordered with sub concepts
o concepts are created through the process of categorisation – mentally
grouping objects together based on common properties
- The level we categorise objects at (e.g basic, subordinate) obviously depends on our
knowledge of it. This also influences the knowledge we can access about it / the
interferences we can make about it
- Without conceptual knowledge / categorisation, you would struggle everyday with
existence
o For every new object encountered (even if similar ones were seen before)
you would be unable to recognise/categorise it and make inferences about it
The definitional approach: looking for defining features
- Concepts have lists of defining features associated with them. Objects with those
features are categorised as being an instance of that concept
- Some concepts are well defined, they have defining features that set them apart
fo othe oepts e.g. dogs ak, ats dot
- Using defining features would work for those objects
- Pole: soe ojets dot hae all defiig featues assoiated ith a oept ut
are still categorised as being an instance of it
The prototype approach: comparing to an average category member
- According to Rosch (1973), categorisation is determined by comparing an object to a
prototype that representing the concept and determining if there a match
- A pototpe is a etal epesetatio of hat a tpial atego ee is like,
ad is ased o a aeage of ool epeieed category members.
- Some objects within a category (e.g fruit) more closely resemble the prototype than
others. The prototypical items are usually named first when listing category
exemplars (Mervis, et al, 1976)
- How closely an object resembles a category prototype influences how quickly it is
categorised
- We likely use prototypes and features for categorisation
- If an object closely matches a prototype, it is quickly categorised
o Example: is the object on the right a pen?
o It matches the prototype well, so yes
- If a ojet does ot esele a pototpe, it aot e uikl ategoised. Its
features are likely scrutinised / it is processed less quickly.
How is Conceptual Knowledge Stored in the Mind?
- So far we have considered how concepts are created via categorisation.
find more resources at oneclass.com
find more resources at oneclass.com
- We know consider how conceptual information (i.e object properties) is stored in
the mind
- Two theories
o Collins and Quillans (1969) Hierarchal network Model of Semantic Memory
o The Connectionist Approach (this is the general theory of mental processing
and does not just explain how conceptual knowledge is stored)
- Both have their roots in the idea that the brain is like a computer (i.e it has limited
processing power / storage) and now use computer-inspired language to describe
how conceptual knowledge is stored (e.g. nodes, networks, inputs etc)
Collins & Quillians (1969) Hierarchal Network Model of Sematic Memory
- We have seen that concepts can be arranged in hierarchies, from superordinate
levels (at the top) to subordinate levels (at the bottom)
- The sematic network approach argues that conceptual knowledge is also arranged
hierarchically, with generic object properties (e.g information genertic to all
mammals) stored higher in the hierarchy than object specific properties (e.g
information specific to canary)
Sematic Network Approach
- The bottom of the hierarchy has specific concepts such as CANARY
o Properties specific to these concepts are stored here (e.g. can sing, is yellow)
- Higher up the hierarchy are more general concepts (BIRD, ANIMAL)
o Properties specific to these concepts are stored here (e.g all animals eat and
breathe)
- The links connecting concepts show they are related to each other in the mind (e.g
there is an association between canary and a bird)
- Each concept (potentionally) has four relation statements linking to propertites
o Is a A aa is a id
o is A aa is ello
o a a id a fl
o has a aa has igs
- Accessing property information: we see and recognise a canary. We first access the
canary concept and properties specific to that concept (it is yellow, it can sing)
- To access more information we must move up a level in the network. Here we access
the BIRD concept and the properties specific to that concept (e.g they have feathers,
they can fly)
- Cognitive economy makes storage efficient, but problems can occur is properties at a
highe oept leel e.g a fl at the BI‘D leel ae ot eleat
- Is the theory correct? If so, the time taken to retrieve information about a concept
(e.g if a canary is a bird) should be determined by the distance that it must be
travelled through the network
- Collins and Quillian (1969) found evidence supporting this prediction using a
sentence verification task
o Eaple: it took loge to eif a aa is a aial tha to eif a
aa is a id
- The odel pedits that A pig is an animal ould e eified oe slol as e
find more resources at oneclass.com
find more resources at oneclass.com
Document Summary
The level we categorise objects at (e. g basic, subordinate) obviously depends on our knowledge of it. This also influences the knowledge we can access about it / the interferences we can make about it. Without conceptual knowledge / categorisation, you would struggle everyday with existence: for every new object encountered (even if similar ones were seen before) you would be unable to recognise/categorise it and make inferences about it. Concepts have lists of defining features associated with them. Objects with those features are categorised as being an instance of that concept. Some concepts are well defined, they have defining features that set them apart f(cid:396)o(cid:373) othe(cid:396) (cid:272)o(cid:374)(cid:272)epts (cid:894)e. g. dogs (cid:271)a(cid:396)k, (cid:272)ats do(cid:374)(cid:859)t(cid:895) Using defining features would work for those objects. P(cid:396)o(cid:271)le(cid:373): so(cid:373)e o(cid:271)je(cid:272)ts do(cid:374)(cid:859)t ha(cid:448)e all defi(cid:374)i(cid:374)g featu(cid:396)es asso(cid:272)iated (cid:449)ith a (cid:272)o(cid:374)(cid:272)ept (cid:271)ut are still categorised as being an instance of it. The prototype approach: comparing to an average category member.
