A.M. Rusanen

advertisement
Anna-Mari Rusanen
Department of Philosophy, History,
Culture and Art Studies
University of Helsinki
Department of Physics
University of Helsinki
There are hundreds (thousands?) studies, which
indicates the process of conceptual change is one of
the key charasteristics of science learning
 Even within the devoted literature on conceptual
change, there is no agreement on how to explain
conceptual change


By conceptual change cognitive scientists and
cognitive psychologists mean (roughly) a specific
kind of learning process:
 a student does not merely accumulate more
knowledge,
 but her conceptions of phenomena in a certain
domain undergo a restructuring process
 that affects ontological commitments, inferential
relations, and standards of explanation.
State 1
can be characterized:
as transformation of the initial
knowledge-state ( for example, a
commonsense picture of the world)
to one of various outcome knowledge
states.
 The outcome:


 a scientific conception (when the learning
process has been successful)
 or one of a number of unscientific
misconceptions (when it has not).

State 2

There are different forms of conceptual change, for
example:
1)Revision:
In some cases cc requires a revision of existing
conceptual system
For example:
▪ Category shifts (Chi)
▪ Tree jumping (Thagard)
▪ Intergration of a conceptual system (diSessa)
2) Reinterpretation:
In some cases cc requires that a learner gives a new
interpretation for the existing concepts/conceptual
system
For example:
▪ Resubsumption (Ohlsson)
▪ Differentiation and coalescence (Carey)
3) Invention:
In some cases cc requires construction or production
of novel conceptual systems
 Even in a way that makes the new and old
systems ”incommensurable” (Carey)
 For example: bootstrapping
4) And so on…
Input


Outco
me
Typically involves:
1) analysing/characterizing the specific cognitive
task (=information processing task) being
performed by a system
2) describing how a certain cognitive mechanism
executes/produces/sustains the phenomenon
▪ Often requires the decomposition of a complex
mechanisms into simpler ones
▪ Is typically given by specifying the precise algorithms
3) describing, how the mechanism is implemented
The description for the task
is given by characterizing
the information processing
task:
 What is the specific
cognitive task
(=information
processing task) being
performed by a system?
 Why it needs to be
performed?
TASK?
Input

Outco
me
The characterization of
information processing
task creates also some
constraints for the
possible underlying
mechanisms:
 Characterizes, why
certain (but not all!)
learning mechanism are
appropriate for fulfilling
the cognitive task.
TASK
Input

Outco
me
Input

?
not often clearly addressed in the literature, but
many share the same intuition:
 Reorganization of the conceptual system, which (in
a case of succesful learning) for example
▪ Makes the system more fruitful, intelligible etc.
(Posner & Strike)
▪ makes the conceptual system more useful
(Ohlsson)
▪ integrates the piecemeal structure of a conceptual
system, and makes the system more coherent
(Disessa, Thagard)

Input

?
Outco
me
How is the task executed/ performed?
 How are the inputs and outputs represented in the
mechanism?
 How does the mechanism transform the input to
generate the output (step by step )?
 Often requires the decomposition of complex
mechanism into simpler ones
Several attempts to define,
but broadly (Bechtel):
 A mechanism is a structure
performing a (information
processing) function in

 virtue of its components parts,
 component operations,
 and their organization.

A description of a
mechanism should describe
this organisation in a detail
Many suggestions for the cognitive ”mechanisms”
of conceptual change
 Many characterizations for the task; many
suggestions for the mechanisms
 Two major category:
 Revision- mechanisms: Mechanisms underlying
conceptual change revise the conceptual system
by changing its conceptual organisation
 Production of new concepts: Mechanisms
underlying conceptual change produce novel
concepts/conceptual systems

”MECHANISMS” of CONCEPTUAL CHANGE
PRODUCTION
of new
concepts by
REVISION by
assimilation &
accomodation
(Vosniadou)
integration &
reorganisation
Concept combination
(with combinatorial
syntax)
(diSessa)
Bootstrapping
Category shifting
& recategorisation
resubsumption
(Chi)
(Ohlsson)
Differentiation and
Coalescence
(Carey)
(Carey)
Often the descriptions of these ”mechanisms of
conceptual change” are quite shallow and offer no
information about the precise structure of
mechanisms/how they work
 For example, Chi et al:
 ”Conceptual change is the process of removing
misconceptions… (which) are, in fact,
miscategorizations of concepts”
 and so ”conceptual change is merely a process of
reassigning or shifting a miscategorized concept
from one category to another”





Chi: By assimilation (=adding
new information) and
recategorization
What is categorizing/
recategorization?
“[c]ategorizing is the process
of identifying or assigning a
concept to category to which it
belongs“ (Chi 2008, 62).
Chi offers no description, how
these identifyings/assignings
are supposed to happen
State
1

Recategorization
The description ?
State
2

If the structure of a mechanism is not specified, a
description offers (at its best) a sketch for a possible
mechanism, not a suggestion for an explanatory
mechanism
 An explanatory model of a mechanism requires a
sufficiently detailed and accurate description (Bechtel
MDC, 2000; Craver, 2006/2007)
When the details of these mechanisms
(reorganisation, bootstrapping, resubsumption,
category shifts, etc.) is analyzed:
 they are often just collections of some more basic
cognitive mechanisms, which are ultimately
responsible for the conceptual change:

▪ Categorization, mapping, transfer, assimilation,
accomodation, analogical reasoning, inductive inference,
abduction…





Imagine, you´d have to learn a novel concept
”Cognitive architecture”
How do you learn it?
Carey: by building a model of the target
How do you do it?
A learning mechanism: By bootstrapping i.e. by using
some of your existing concepts to build a new concept
1) Initial State:
 occurs when a learner encounters a set of
interrelated explicit symbols, such as symbols of a
scientific theory
 ”Cognitive architecture, compositionality,
information semantics, representation…”
 These symbols, ”PLACEHOLDERS”, are
uninterpreted:
 Are partially mapped/not mapped into any
already existing concepts
2) The process of conceptual change:
These placeholders are then taken up by various
”modeling processes”, which include
 Inductive reasoning, analogical reasoning,
abduction etc.
 If you know something about models in
physics, you might think that the structure of
”cognitive architecture” is analogical
 These cognitive processes ”produce” the content
for placeholders
3) The Outcome:
When these placeholder symbols have a stable
conceptual role in a new structure, they have a
conceptual content in virtue of their conceptual role
 The new structure may be incommensurable with
the old one


When ”bootstrapping” is given a description, it turns
out to be a collection of some more ”basic”
mechanisms:
 Mental modeling (?), analogical reasoning,
inductive reasoning, abduction, etc.

Ohlsson describes the process of resubsumption:
 Conceptual change happens, when a person uses
analogical transfer to map conceptual system
from one domain A to a new domain B, which has
been earlier conceptualized by another system
 If the new system is evaluated to be more useful,
the target domain is reinterpreted by it
 The mechanisms: analogy, transfer, mapping,
interpretation…

There is evidence that for example analogical
reasoning (and transfer) play a crucial role in some
forms of conceptual change
 BUT: not much is known, how, for example,
analogical reasoning should be understood in the
context of complex learning in adults (Markman)

In addition, from the cognitive point of view,
analogical reasoning is a complex process, and may
involve several other mechanisms:
 Similarity comparisons, visuo-spatial mechanisms
for imagination, mechanisms for mapping from
one system to another…
 These mechanisms may involve several
submechanisms
▪ there is evidence that even some motor controlmechanisms are used in some sort of mappings
(number learning in childhood…?)

Requires that the mechanisms responsible for a
certain type of conceptual change should be
specified in a detail
 Can be really challenging in the case of conceptual
change, because it may involve a hierarchical
collection of many different submechanisms
 Some of those are better ”known”
(categorization, inductive reasoning), some of
those aren´t (mapping mechanisms?)

In addition, there are many different forms of
conceptual change, and they may involve several
different mechanisms
 Ohlsson, 2009: A theory of cc cannot just be the
list of all possible mechanisms, it must also
constraint mechanisms
 It must be able to tell, why certain mechanisms
are appropriate for cc, and why some other aren´t
 For this reason, the task level also matters
From a cognitive science point of view, explaining
conceptual change requires:
1) A precise description for the information
processing task
2) A sufficiently accurate and detailed description of
the mechanisms responsible for the task
 can be a really challenging task!



But why to bother? Do we really need to know?
Yes:
 If you do not know, how it works, you do not
know, how to manipulate it
 If you do not know, how to manipulate it, you
don´t know, how to facilitate it
Download