Paper: T.R.G. Green (1989) Cognitive dimensions of notations. In A.Sutcliffe and L. Macaulay (Eds.) People and computers V. Cambridge UK: Cambridge University Press: 443-460.
Paper selected and presented by Martin Oliver.
(Notes below from Carey Jewitt – please add to these)
Context for the paper
Martin contextualised the paper for the group. It was a starting point, an early paper, on cognitive dimensions, work that has been developed further by the author and others in the field of cognitive science. (Sara Price has some interesting suggestions for futher readings that show how this work has been developed – Maybe you could post the links Sara?). Martin also sketched a sense of the disciplinary context that the paper comes from: the thinking through of metaphors as ways to explore and explain, predict and model how people act in the world, One purpose for doing this is to tryout, develop and test, tentative ‘laws’ (laws in a soft sense) for understanding how people think – a ‘big work in progress’ agenda.
More specifically this paper is part of Green’s work in the psychology of programming group which asked, among other questions, why can it be so hard to learn programming. This paper marks a move away from talking about the BEST programming language to thinking about fitness for purpose – in which each programming language would offer losses and gains which need to be understood in the context of use. In short a move away from a focus on the representational system per se to the system in use. Here Martin drew a parallel with linguistics and the study of Langue as opposed to Parole – this was picked up later in our discussion about the possibility (or usefulness of thinking of the possibility?) of separating system, notation, user, and environment - and the problem that would arise for cognitive dimensions of allowing the user to be too complex in a process that is about trying to define and predict (control?) their behaviour.
The question for us as a reading group seemed to be, ‘In what ways might the cognitive dimensions that the paper sketches be useful for helping us think about design, learning and pedagogy in our different research contexts?’ The paper produced an interesting discussion.
Cognitive dimensions?
According to Green, the cognitive dimensions of a notation are:
• A characteristic of the way that info is structured and represented
• One that is shared by many notations of different types
• And by its interaction with the human cognitive architecture has a strong influence on how people use the notation
• And affects whether the strategy of opportunistic planning can be pursued.
We discussed – where do these cognitive dimensions ‘come from’ – is it a model or a theory – or common sense from observations? Green appears to have developed these on the back of observational studies and the literature as one way to explain how people interact and act.
Green draws in the social context of programing by stating that ‘system=notation+ environment’ (pg 2). In our discussion there was a general sense that the differences between people are underplayed and the ways in which users/people change via their interaction with a system is not accounted for.
Some key points from our discussion of paper
here I focus on 3 of the dimensions that we discussed and seemed – in my opinion - to have most currency with the group
7.1 Hidden/explicit dependencies
This looks at the balance between what is hidden and explicit in an application (e.g. in a spreadsheet application) and what this design decision means for how people can interact with an application. In the spreadsheet example given in the paper, the higher the level of hidden dependencies between cells in the spreadsheet the more difficult it is to ensure the accuracy of data when cells are modified because the dependencies between cells are not easily recoverable. On the other hand if the algebraic model is explicit it increases the amount of data/info to deal with and this may decrease the ability of the user to recognise salience.
The design implication here is that all the information necessary/that matters to the user for what it is that they need to be able to do should be accessible and environmental support )cross references, browsers, etc) given.
7.2 Viscosity/fluidity
This dimension is concerned with the extent to which something – in this case a notation system - resists local change. We discussed the payoff here between safeguards and easiness in changing something at a local level, and the different contexts where these are useful.
Implications – viscous systems create more work for the user, - breaking track of thought and creating ‘working memory load’ . creating modules is one way of managing a viscous notation system, parcelling up aspects. There may be good pedagogic reasons for making things difficult to do. Viscosity might demand serious thinking before action – rather than evolutionary thinking…
7.3 Premature commitment
Here the focus was on generative ordering – the demands made by an application: whether a program demands that the user write in any order or in specific final text order – the extent to which the user has to look ahead imposes different degrees of commitments… this has implications for dealing with contingency of design as re-design.
Implications - The possibility of decoupling the generative order from the final text order has implications for how designs can be developed
General points re cognitive dimensions and design
Other cognitive dimensions noted in the paper included – role expressiveness, diffuseness, consistency, susceptibility, perceptual cues, cognitive off-loading.
The general sense of the group was, I think that even if the dimensions turn out not to be ‘true’ they nonetheless give a useful terminology for talking together about an aspect of a design – that they useful tools.
Cognitive dimensions appeared to offer ways of understanding the mindset of a designer – in order to understand a program/ application. They also offer a way to see how the designer/design imagines (wants to produce?) the learner/user. One point raised here is how the user is usually imagined as an individual, but this is questioned by the example of ‘extreme programming’ (where programmers work in pairs – one coding, the other checking, and alternating this role – as a way to reduce mistakes)
The question of where might the design deliberately ‘stop’ the user from acting– e.g. via the designed viscousity of the program, or what relationships are made apparenet – explicit and which hidden might be a marker of what is valued, and the expectations etc by the designer. The possibilities to customise what is hidden etc (values) was discussed in pedagogic terms.
The question ‘who decides what representation to start with?’ was asked - why talk as opposed to written, or visual , or audio? This prompted a brief discussion of the histories of representational forms, the cultural and social histories of notations and the ways in which these create positions to knowledge, boundary disciplines, create and gateway professions, It led to the question of whether meaning can exist outside of a sign and how usefully think about the relationship between sign and meaning – sign and concept, external and internal signs. This issue will be the focus of our next paper – to be announced shortly.
Tuesday, May 15, 2007
Monday, May 14, 2007
Clash
Ironically I cannot make the reading group this week because there is a clash with an ICN seminar on 'synaesthesia for finger counting and dice patterns' - three different aspects of representation of abstract concepts. There was no mention in Carey's summary of the paper that referred to embodied cognition I think (can't figure out how to look at the rest of the blog while doing this). I'm hoping to find out about the use of fingers and patterns as mediating technologies for coming to an understanding of number - and the role of synasethesia in that - and would like to see this being able to fit into that schema somewhere. Will report back after the seminar.
Friday, May 11, 2007
Session 2: External Cognition (17 may)
Paper: Mike Scaife and Yvonne Rogers (1996) External cognition: how do graphical representations work?
Int. J. Human -Computer Studies 45: 185-213.
Some notes:
1. Internal and external visual representations
• What is the nature of the relationship/ interaction between these? External cognition approach
– External reps a central functional role in relation to internal cognitive mechanisms (not an equivalence structure)
– Focus on the cognitive processing
– The properties of internal and external structures and cognitive benefits of reps
–
2. Three central characteristics for analytical framework
• Computational offloading
• Re-representation
• Graphical constraining (limiting abstraction, restricting interpretation)
3. Resemblance fallacy
• How subjects identify key features and constraints?
• External and internal reps are simply assumed to have same characteristics!
• More a matter of learning to read the external?
4. Processing and the internal/external
• How the I and the E act in concert?
• How understanding is integrated with existing knowledge, interpreted, and re-represented externally?
• Complex, cyclical, interacting processes over time
5. Aspects of design - bridging
• Visual organization structures
• Type of notation (symbols and icons) used
• Canonical forms of representation (e.g. diagrams)
• Recognition of reading rules - cues appropriate inferences
• Ability to use representations
• Degree of abstraction
• Cognitive traces - annotations
• Levels of activity- how people interact with External representation
Int. J. Human -Computer Studies 45: 185-213.
Some notes:
1. Internal and external visual representations
• What is the nature of the relationship/ interaction between these? External cognition approach
– External reps a central functional role in relation to internal cognitive mechanisms (not an equivalence structure)
– Focus on the cognitive processing
– The properties of internal and external structures and cognitive benefits of reps
–
2. Three central characteristics for analytical framework
• Computational offloading
• Re-representation
• Graphical constraining (limiting abstraction, restricting interpretation)
3. Resemblance fallacy
• How subjects identify key features and constraints?
• External and internal reps are simply assumed to have same characteristics!
• More a matter of learning to read the external?
4. Processing and the internal/external
• How the I and the E act in concert?
• How understanding is integrated with existing knowledge, interpreted, and re-represented externally?
• Complex, cyclical, interacting processes over time
5. Aspects of design - bridging
• Visual organization structures
• Type of notation (symbols and icons) used
• Canonical forms of representation (e.g. diagrams)
• Recognition of reading rules - cues appropriate inferences
• Ability to use representations
• Degree of abstraction
• Cognitive traces - annotations
• Levels of activity- how people interact with External representation
Friday, May 4, 2007
Welcome to the reading group
Meetings on the 3rd Tuesday of each month, 12.30 – 1.30 LKL small seminar room
For copy of the reading please email Elizabeth Andrew.
General information
A broad focus for each session will be to explore the notion of representation with reference to technology and learning. Papers and themes will be suggested by participants of the reading group. Key questions and themes generated in the first session that will inform our discussions include:
• How can the right representation for a learning outcome be identified?
• The design dimensions of representations for learning
• The relationship between cognitive load and cognitive effort
• Designing cognitive or computational constraints
• Designing thinking (problem/search) spaces that learners can benefit from
• The role of abstraction
• The authorship and ownership of representations (the relationship between personal experience and new experiences
• Different forms and modes of representation
• Processes of production of representations
Format
Introduction of the paper to include why chosen, what area related to, what question the authors trying to address. This is followed by informal open discussion
Reading group blog
(you are there) To post notes on each paper, continue the dialogue as needed; to maintain a record of the group; invite the authors whose work we discuss to contribute.
Discussions will be audio recorded.
Session 1 (20 March 2007)
Paper: Donald Peterson (1994) Re-representation and emergent information in three cases of problem solving, in T.Dartnall (ed.) Artificial Intelligence and Creativity, 81-92, Netherlands: Kluwer Academic Publishers.
Presented by Kevin Walker
The chapter appears in a book on creativity and AI. It contains three case studies, in which the author re-represents problems in order to add emergent information which helps solve the problems. A re-representation is not a mere re-description of a problem, he says, unless it produces such emergent information.
The three examples are mathematical puzzles. In each case the author removes unneeded data, puts data into a table then a new representation, in each case 'mathematising' or 'computationalising' the problem. This is important because in the context, the goal is for computers, not humans, to solve the problems. Thus the representations tend to be linear, logical, efficient and mathematical.
Key points from the discussion
The point of the paper was discussed for some time – was it suggesting a way of learning or a way of understanding how the solution process can be expressed as a logical set of rules for computation. This raised questions about the process of producing representations in solving problems: Should these be generated by learners themselves, or is moving through the process of how solutions are generated enough? The question of effective learning was discussed – how much is effective learning the solution of the problem posed, or the ability to generate new problems?
This relates to the level of activity of the learner: Much of our research at LKL centres on learners actively constructing knowledge (at the least) or things (real or virtual) as a means of building knowledge.
The relationship between internal representations (mental models) and external representations was raised. The extent to which abstraction is useful was discussed – with focus on the chess example in the paper it was pointed out that the removal of the example from the context of chess playing did not acknowledge the way in which grandmasters work with such problems – that is, they don’t think like computers but rather their expertise comes from seeing so many board configurations over time; in other words their expertise is tied to the representation of the chess board (see Sweller et al 1998).
Related to this point is the personalised nature of representations. Different learners may find alternative representations useful, but when it comes to sharing knowledge, or participating in a community with particular definitions, common representations (e.g. mathematical equations) are necessary. Language, therefore, is an important bridge between individual and shared representations. All of the examples in the paper are described using narrative means, with supporting illustrations.
What kinds of spaces do the different representations and re-presentations present for the learner was another thread of discussion. The need to reduce the ‘search space’ was discussed. The ways in which microworlds and other forms of constraint are needed to give freedom to explore through constraint.
This led to discussion of what kind of constraints are productive, when and where. Which led onto the notion of cognitive load, overload, off-loading and the relationship between these and learning. When and what is it useful to off-load and what role does/might technology have in this process (e.g. calculators). The question of what is actually being off-loaded and the assumption that this means learning is happening was questioned. Maximum cognitive offload = minimum cognitive effort (= learning?)
The work of other authors brought into the discussion: Scaife and Rogers, Cox,
Green and Blackwell, Petrie and Green, Simon Holland.
References
Sweller, J., van Merriënboer, J.J.G. and Paas, F. (1998) Cognitive architecture and instructional design. Educational Psychology Review 10(3), pp. 251-295.
For copy of the reading please email Elizabeth Andrew.
General information
A broad focus for each session will be to explore the notion of representation with reference to technology and learning. Papers and themes will be suggested by participants of the reading group. Key questions and themes generated in the first session that will inform our discussions include:
• How can the right representation for a learning outcome be identified?
• The design dimensions of representations for learning
• The relationship between cognitive load and cognitive effort
• Designing cognitive or computational constraints
• Designing thinking (problem/search) spaces that learners can benefit from
• The role of abstraction
• The authorship and ownership of representations (the relationship between personal experience and new experiences
• Different forms and modes of representation
• Processes of production of representations
Format
Introduction of the paper to include why chosen, what area related to, what question the authors trying to address. This is followed by informal open discussion
Reading group blog
(you are there) To post notes on each paper, continue the dialogue as needed; to maintain a record of the group; invite the authors whose work we discuss to contribute.
Discussions will be audio recorded.
Session 1 (20 March 2007)
Paper: Donald Peterson (1994) Re-representation and emergent information in three cases of problem solving, in T.Dartnall (ed.) Artificial Intelligence and Creativity, 81-92, Netherlands: Kluwer Academic Publishers.
Presented by Kevin Walker
The chapter appears in a book on creativity and AI. It contains three case studies, in which the author re-represents problems in order to add emergent information which helps solve the problems. A re-representation is not a mere re-description of a problem, he says, unless it produces such emergent information.
The three examples are mathematical puzzles. In each case the author removes unneeded data, puts data into a table then a new representation, in each case 'mathematising' or 'computationalising' the problem. This is important because in the context, the goal is for computers, not humans, to solve the problems. Thus the representations tend to be linear, logical, efficient and mathematical.
Key points from the discussion
The point of the paper was discussed for some time – was it suggesting a way of learning or a way of understanding how the solution process can be expressed as a logical set of rules for computation. This raised questions about the process of producing representations in solving problems: Should these be generated by learners themselves, or is moving through the process of how solutions are generated enough? The question of effective learning was discussed – how much is effective learning the solution of the problem posed, or the ability to generate new problems?
This relates to the level of activity of the learner: Much of our research at LKL centres on learners actively constructing knowledge (at the least) or things (real or virtual) as a means of building knowledge.
The relationship between internal representations (mental models) and external representations was raised. The extent to which abstraction is useful was discussed – with focus on the chess example in the paper it was pointed out that the removal of the example from the context of chess playing did not acknowledge the way in which grandmasters work with such problems – that is, they don’t think like computers but rather their expertise comes from seeing so many board configurations over time; in other words their expertise is tied to the representation of the chess board (see Sweller et al 1998).
Related to this point is the personalised nature of representations. Different learners may find alternative representations useful, but when it comes to sharing knowledge, or participating in a community with particular definitions, common representations (e.g. mathematical equations) are necessary. Language, therefore, is an important bridge between individual and shared representations. All of the examples in the paper are described using narrative means, with supporting illustrations.
What kinds of spaces do the different representations and re-presentations present for the learner was another thread of discussion. The need to reduce the ‘search space’ was discussed. The ways in which microworlds and other forms of constraint are needed to give freedom to explore through constraint.
This led to discussion of what kind of constraints are productive, when and where. Which led onto the notion of cognitive load, overload, off-loading and the relationship between these and learning. When and what is it useful to off-load and what role does/might technology have in this process (e.g. calculators). The question of what is actually being off-loaded and the assumption that this means learning is happening was questioned. Maximum cognitive offload = minimum cognitive effort (= learning?)
The work of other authors brought into the discussion: Scaife and Rogers, Cox,
Green and Blackwell, Petrie and Green, Simon Holland.
References
Sweller, J., van Merriënboer, J.J.G. and Paas, F. (1998) Cognitive architecture and instructional design. Educational Psychology Review 10(3), pp. 251-295.
Subscribe to:
Posts (Atom)