Cognitive models of designing

1
Cognitive models of designing

• Willemien VISSER
• INRIA - National Research Institute for Computer
  Science and Control
• EIFFEL team - Cognition Cooperation in Design
• Rocquencourt - 78153 Le Chesnay Cedex (France)
• email Willemien.Visser_at_inria.fr
• DESC9099 How designers think
• September 2004

2
Outline

• A descriptive, cognitive ergonomics approach to
  design grounded in cognitive psychology
• The two main paradigms in the descriptive,
  cognitive approach to design
• Design as a particular type of problem main
  characteristics
• Design as a particular type of activity main
  characteristics

3
A descriptive, cognitive ergonomics approach to
design grounded in cognitive psychology

• a cognitive ergonomics approach to design
• a descriptive approach to design
• an approach to design grounded in cognitive
  psychology
• The two main paradigms in the descriptive
  approach to design

4
A cognitive ergonomics approach

• definitions by the International Ergonomics
  Association (IEA)
• "Ergonomics (or human factors) is the scientific
  discipline concerned with the understanding of
  interactions among humans and other elements of a
  system, and the profession that applies theory,
  principles, data and methods to design in order
  to optimize human well-being and overall system
  performance. Ergonomists contribute to the design
  and evaluation of tasks, jobs, products,
  environments and systems in order to make them
  compatible with the needs, abilities and
  limitations of people."
• "Cognitive ergonomics is concerned with mental
  processes, such as perception, memory, reasoning,
  and motor response, as they affect interactions
  among humans and other elements of a system."

5
A cognitive ergonomics approach

• since 1980, cognitive design studies in
  ergonomics
• two objectives knowledge (epistemic) and
  assistance
• these objectives are related specifications for
  design assistance require models of the
  cognitive, underlying processes actually used by
  designers

6
"Design" a global definition from a cognitive
viewpoint

• specifying an artefact, given requirements that
  indicate one or more functions to be fulfilled
  and/or objectives to be satisfied by the artefact
• from a cognitive-activity point of view
• to elaborate representations (mental and
  external) of the to-be-specified artefact until
  they are so precise and detailed that they
  completely specify the artefact's implementation
• NOTE requirements development ? design and
  implementation ? design
• however, both may require decisions involving
  "design"
• cf. "the intermingled character of design and
  implementation", but also of requirements
  development and design!

7
A descriptive approach to design

• vs. prescriptive approach to design generally,
  design methodologies
• linear, sequential design process
• following an abstract-concrete axis conceptual ?
  physical specs
• generally based on stage models iteration of
  stages to be followed in a stepwise, top-down
  manner
• vs. process models activities actually carried
  out
• descriptive models are generally process models
• existing descriptive models proposed in cognitive
  ergonomics focus on a particular component of
  design
• Darses constraint management (esp. design
  solution generation)
• Bonnardel design solution evaluation
• Burkhardt and Détienne Visser reuse in design
• Visser planning and organisation of design
  activity
• NOTE no predictive models yet

8
An approach grounded in cognitive psychology (1)

• "cognitive" aspects of design
• cognitive processes
• knowledge and representations
• NOTE "representation" here is a "mental
  representation" resulting from an interpretation
  activity
• this presentation
• cognitive processes
• knowledge and representations actually used by
  designers
• dynamic cognitive aspects
• focus on design activity, rather than
• design task
• results of the activity

9
Task vs. Activity

• task
• goal goal-achievement conditions
• a "unit" of work
• prescribed vs. actual task
• "prescribed" task what people are supposed to do
• "actual" task what people assign themselves
• activity
• the way people actually realise their task
• cognitive activity
• the way people actually realise their task on a
  cognitive level

10
An approach grounded in cognitive psychology (2)

• starting point the classical cognitivist
  paradigm
• design analysed as problem solving Herbert
  Simon The sciences of the artificial, 1969
• my view
• design involves problem solving,
• BUT design is not "problem solving"
• design is a construction of representations
• (representations specifications for the
  artefact)

11
Individual design

• until 1990, cognitive design studies concerned
  individual design
• design is most often teamwork, but
• an important proportion of work on design
  projects is individual
• collective design involves the cognitive
  activities implemented in individual design, in
  addition to activities specific to collective
  work
• development of appropriate work environments
  requires analysis of the articulation between
  reasoning implemented in individually and in
  collectively conducted activities

12
"Generic design" hypothesis

• underlying conjecture
• important similarities between design tasks in
  different domains (e.g. architecture, mechanical
  design, software design, but also traffic signal
  setting and planning of routes or meals)
• and
• important differences between design tasks and
  non-design tasks
• but
• there are different "forms" of design

13
A descriptive, cognitive ergonomics approach to
design grounded in cognitive psychology summary

• the focus on activity has consequences for
  data-collection methods a privileged method is
• real-time observation
• preferentially in the designers' workplace
  ("field studies")
• underlying data
• observational field studies on professional
  designers working in industrial design contexts
• domains mechanical, industrial and software
  design
• this presentation a characterisation of design
  based on empirical data concerning the actual
  design activity implemented by individual
  designers during their work on professional
  design projects (vs. data on the design task,
  process or results)

14
The two main paradigms in the descriptive,
cognitive approach to design

• the classical symbolic information-processing
  (SIP) approach to design (the "cognitivistic",
  "computational" view) Herbert A. Simon
• the situativity (SIT) approach to design
  "situated action", "situated cognition" Donald
  Schön, Louis Bucciarelli, John Gero

15
My view on SIP vs. SIT

• SIT dominant resource is designers' situational
  context
• SIP dominant resource are designers'
  problem-solving methods
• SIP central role of designers' knowledge and
  internal representations SIP does not deny that
  these may evolve under the influence of
  designer's activity (I.e. interaction with their
  environment)nevertheless, SIP does not give it
  much attention
• SIT central role of designers' environmentSIT
  does not deny the existence of internal
  resourceshowever, SIT does not give it much
  attention

16
My approach in relation to both SIP and SIT

• for a designer confronted with a design task, all
  resources for action depend on both internal
  (knowledge, representation) and external (social
  and artefactual) data and their relationship
• however, these two types of data are not in a
  symmetric relation it's the designer who, using
  their knowledge and interpretation activities,
  establishes the relationship between internal and
  external data
• central role of construction of representations!

17
Design as a problem - Design as an activity

• Two complementary viewpoints on design
• Design as a problem design projects constitute
  problems in that they are tasks for which
  designers cannot evoke, from memory, a procedure
  allowing to attain their goal (I.e. to specify
  the to-be-designed artefact)
• Design as an activity in order to specify the
  to-be-designed artefact, designers will have an
  activity that has many cognitive aspects
• main characteristics of these two viewpoints on
  design

18
Design as a problem

• Some preliminary definitions
• "problem" and " "problem-solving"
• designing as nonroutine-problem solving
• problem solving as a global process
• design involves problem solving,
• BUT design is not "problem solving"
• Design as a particular type of problem main
  characteristics

19
Design as a problem

• "problem"
• pre-scientific, common-sense meaning
  "difficulty", "deficiency"
• here specific, technical sense used in cognitive
  psychology
• a person's representation of their task that does
  not immediately evoke a" legal" procedure
  allowing to attain the goal (Newell Simon,
  1972)
• problem-solving tasks vs. execution tasks
• problem-solving elaboration of procedures
• execution applying a previously elaborated
  procedure
• "problem" character of a task relative feature,
  depending on
• task situation
• person confronted with the task (esp. their
  knowledge, experience)

20
Design as a problem designing as
nonroutine-problem solving

• my view design problems are nonroutine problems
• "routine" - "nonroutine" problems distinction
  (Mayer)
• routine problems tasks that, although not
  eliciting a memorized answer, can be solved by
  applying a well-known procedure
• nonroutine problems tasks for which one does not
  have a well-known solution procedure, so that one
  must generate a novel procedure
• ? defined from a cognitive viewpoint, the
  routine-problem solving activities involved in a
  design project are not considered "design"
  activity

21
Design as a problem problem solving as a global
process

• "problem solving" a global process, not a
  particular stage in the design process
• taking place through a big number of small steps
• e.g. "problem forming", "problem finding"
  (Simon), "problem attempting", "problem
  formulation" , "problem setting", "problem
  structuring" (Schön), "problem conceptualising",
  "problem framing" or "reframing" , "problem
  defining" or "redefining"
• cognitive psychology approach to problem solving
  
• "problem solving" refers to all activities that
  lead from a problem specification to its
  solutions
• even if sometimes the term "problem solving" is
  used with a restricted meaning, referring to an
  activity different from, especially, "problem
  forming" or " problem structuring"

22
Design involves problem solving, BUT design is
not "problem solving"

• cf. Design as an activity
• construction of representations
• but also (not in this presentation)
• socio-cultural aspects (esp. in collective
  design)
• emotional aspects

23
Design as a particular type of problem main
characteristics

• ill-definedness
• ambiguity
• complexity
• constrained character of design solutions
• several more or less satisfactory, rather than
  unique correct solutions
• solutions with impact on people and their
  environment

24
Ill-definedness

• design problems' ill-definedness concerns all
  three problem-solving components initial state,
  goal state and operators (Reitman)
• in design problems, generally, only the goal
  state receives some explicit specification, even
  if poorly and at an abstract level
• by way of the artefact's function and goals, and
  of constraints on the artefact
• initial state and operators are generally
  underspecified
• problem definedness is a relative characteristic
• related notions "wicked" (vs. "tame" or
  "benign", Rittel and Webber), "open-ended",
  "ambiguous", "vague", "real-world"
• ill-defined problem solving is not only a
  question of structuring (Simon's view)

25
Ambiguity

• people differ with respect to their
  interpretation of the problem (its states and the
  operators that are "legal")
• related to ill-definedness (Reitman)
• poor agreement regarding referents of problem
  components and other problem characteristics
• ? solutions to ill-defined problems are only
  "accepted" and only "more or less" so (even in a
  specified community)

26
Complexity

• design problems are generally large and complex
• "large" many components (sub-problems) ? long
  duration
• "complex" (cybernetics) information content of a
  group of elements assembled in a system by way of
  relations
• critical element relations between the elements
  and their unforeseeability
• design problems many interdependent components
• ? problem decomposition in order to make
  resolution more feasible
• BUT decomposing design problems generally is not
  evident

27
Constrained character of design solutions

• A.I. models design as constraint management and
  satisfaction
• cognitive ergonomics studies on the actual use of
  constraints (Bonnardel, Darses)
• ? central role in design
• BUT design activity also involves other types of
  knowledge and other types of processes
• design constraints are often conflicting
• ? different trade-offs may require satisficing
• ? several, more or less acceptable solutions
• distinguish generative and critical (evaluation
  criteria) constraints
• constraints are more or less debatable, depending
  on their nature social, political, economic,
  legal

28
Several more or less satisfactory, rather than
unique correct solutions

• design solutions (i.e. specifications for the
  artefact) also are debatable
• ? they may be contested
• by other designers or by the client
• not for being incorrect, but because other
  choices are possible
• the same function or need may be satisfied in
  several ways
• there is no objective ranking (validity, weight)
  for many design criteria
• ? different designers come up with different
  design proposals

29
Solutions with impact on people and on their
environment

• no definite tests for solutions to ill defined
  problems
• BUT consequences of design solutions
• may extend over long periods of time
• may be of concern to many people
• design involves anticipating the artefact's
  impact on human activity (anticipating people's
  future use of artefact)
• by definition, the to-be-designed artefact cannot
  be tested, only its version "under design"
• ? omissions, failures and other complications
  often remain unnoticed until it is "too late"
• possibilities for testing the artefact's
  under-design version depend on the domain of
  design
• e.g. in HCI simulation, use of prototypes

30
Design as a particular type of activity its
main characteristics

• from a cognitive-activity point of view
• given requirements that specify one or more
  functions to be fulfilled and/or objectives to be
  satisfied by an artefact, elaborate
  representations of the artefact until they are so
  precise and detailed that they completely specify
  the artefact's implementation
• ? central role of knowledge and representations
  (mental and external)
• ? design involves problem solving, BUT design
  is not "problem solving"

31
Design as a particular type of activity its
main characteristics

• Design as a type of cognitive activity, rather
  than a professional status
• Design as satisficing, rather than optimising
• Creativity
• Problem-representation construction, solution
  generation and solution evaluation
• Using knowledge from different domains and
  abstraction levels
• Constructing and using different types of
  representations
• Selecting and sticking to a "kernel idea"
• Some design strategies

32
Design as a type of cognitive activity, rather
than a professional status

• already formulated by Simon (1969)
• from a cognitive viewpoint
• many activities of professionals who are not
  qualified as "designers" are considered "design"
  (e.g. many draughtsmen, technicians)
• many activities performed by professionals
  qualified as "designers" are not considered
  "design" ("routine design" or other activities
  such as many public relations or general
  management activities)
• an actvity is not per se "design" or not (e.g.
  drawing or calculating)

33
Design as satisficing, rather than optimising

• introduced by Simon
• (cf. his Nobel prize human bounded rationality
  in economics)
• satisficing is to "settle for the good enough",
  rather than to optimise
• ? making decisions without complete information,
  accepting one or more solutions that are judged
  satisfactory, rather than calculating the
  optimum one
• human bounded rationality explains the need for
  satisficing
• people's limited abilities with respect to both
  memory contents and processing
• satisficing both in solution generation and
  solution evaluation

34
Creativity

• "creative" often qualifies the result of an
  activity ("innovative")
• here "creative" is no synonym of nonroutine
  design problem solving, but a particular
  characteristic of this activity
• "Arriving at a solution by strict calculation is
  not regarded as designing". Yet, "just how much
  originality is needed to distinguish the
  preparation of working details from actual
  designing, and just how little is inevitably
  required to distinguish calculation from
  designing, is very difficult to prescribe."
  (Archer)
• designing differs from artistic creation
  "design" rather than "creation" if
• start from requirements (constraints, functions,
  objectives)
• (aim to) implement the resulting specifications
  as an artefact

35
Problem-representation construction, solution
generation and solution evaluation

• considered from an analytical viewpoint, at a
  high level of abstraction, design may be said to
  proceed through three phases
• considered from another analytical, less
  high-level perspective, design may be said to
  proceed through numerous iterative cycles
• from a globally specified problem (design
  requirements) to a detailed implementable
  solution (artefact's specifications)
• in each cycle, problem solving proceeds in three
  steps
• however, these are analytical models that do not
  render the actual design activity

36
Problem-representation construction, solution
generation and solution evaluation

• in actual design, problem representation,
  solution generation and solution evaluation are
  intertwined
• design is organised opportunistically one factor
  of opportunism is the way in which designers
  handle problem decomposition and refinement
• problems are worked out, simultaneously, in two
  directions
• in breadth decomposing a problem into
  sub-problems
• (requiring integration of intermediate solutions)
• in depth refining sub-problems, on an
  abstract-concrete dimension (until an
  implementable solution is reached)
• (see examples in paper)

37
Using knowledge from different domains and
abstraction levels

• design is a knowledge-intensive problem solving
  activity different types of knowledge
• application domain (one's particular design
  expertise)
• underlying technical domains and theory
• design methods
• non technical domains
• ergonomics (some basic knowledge)
• social, political, economic, or legal aspects of
  the artefact and its use

38
Using knowledge from different domains and
abstraction levels

• designers are generally expert in one particular
  domain
• ? design projects often require collaboration
  between specialists from various domains
• ? large projects are mostly a matter of
  collective design
• knowledge from different abstraction levels
• generic, abstract knowledge
• specific knowledge (reuse)

39
Constructing and using different types of
representations

• both internal, i.e. mental, and external
  representations
• both constructed using knowledge
• external representations
• all through the design process, construction and
  use of external representations
• representations constructed and used in early
  design generally differ from the final
  representation
• according to the domain of design, intermediary
  representations may be textual, graphical,
  three-dimensional
• sketches, flowcharts, plans, scenarios, scale
  models, prototypes

40
Constructing and using different types of
representations

• functions of external, intermediary
  representations
• storing solutions
• communicating
• "computational offloading"
• reasoning (cf. Schön's "reflective conversation
  with the situation")

41
Constructing and using different types of
representations

• "viewpoint"
• notion recently adopted by several authors
• no clear definitions
• no reference to a particular type of
  representation
• often reference to design participants'
  particular professional knowledge, domain of
  expertise and/or know-how
• ? no particular characteristics compared to those
  classically attributed to other forms of
  representations

42
Constructing and using different types of
representations

• representations of different abstraction levels
• from representations related to the artefact's
  purpose
• to representations related to the artefact's
  physical properties
• no systematic, fixed order designers come and go
  between representations at level n and
  representations at level nm

43
Constructing and using different types of
representations

• different representational formats
• e.g., graphical external representations
• e.g., in architectural or mechanical design,
  differences depending on the phase of design
• initially necessarily vague
• explore possibilities without handling with
  details
• allow designers to maintain as many degrees of
  freedom as possible in their solution (cf.
  refraining from premature commitment)

44
Selecting and sticking to a "kernel idea"

• observed in various empirical cognitive design
  studies
• in an early phase of their activity, designers
  tend to select a "kernel idea" and to stick to
  this "central concept" (Ullman, Dietterich, and
  Staufer) in what is going to become their global
  design solution
• "primary generator" (Darke, 1979) a few simple
  objectives adopted at the start of a project in
  order to generate the initial solution kernel
• cf. "fixation"

45
Selecting and sticking to a "kernel idea"

• systematic character of the kernel idea?
• hypothesis such "early fixation" on a kernel
  idea is induced by individual and/or
  de-contextualised design (Ball and Ormerod study
  on "real" design review meetings)
• but also studies of individually conducted design
  have observed designers to come up with several
  solution ideas (Eastman, Whitefield)
• contradiction with designers' avoiding premature
  commitment?
• perhaps only apparent
• designers declarations and/or aspirations may
  remain theory early on in design at a conceptual
  level, select a kernel idea
• but later on at a more concrete level, not fix
  (all) values of its variables

46
Design strategies

• design strategies used by experienced,
  professional designers (cf. "experts" vs.
  "novices")
• decomposition and planning
• planning and organisation
• reuse
• simulation

47
Conclusion

• a descriptive, cognitive ergonomics approach to
  design grounded in cognitive psychology
• design as a problem and design as an activity
• importance of representations external, but
  especially mental
• design involves problem solving, but design is
  not problem solving
• from a cognitive viewpoint, design is essentially
  construction of representations

48
(No Transcript)
49
Complementary material

• My view on SIP vs. SIT - My approach in relation
  to both SIP and SIT details (5 slides)
• Problem-representation construction, solution
  generation and solution evaluation details (3
  slides)
• Design strategies details (13 slides)

50
My view on SIP vs. SIT - My approach in relation
to both SIP and SIT details (5 slides)
51
My view on the classical SIP approach to design
recognition and critique

• my view
• several essential characteristics of design
  identified by Simon still constitute an important
  contribution to cognitive design models
• BUT
• Simon's approach also evokes critique-gt modify
  and complete Simon's frameworkpartly based on
  situativity ideas

52
My view on the classical SIP approach to design
elements of recognition

• the main essential characteristics of design
  identified by Simon that constitute an essential
  contribution to cognitive design models
• ill-structuredness (ill-definedness) of problems
• design as a type of cognitive activity rather
  than as a professional status
• design as "satisficing"
• instead of "design as a problem-solving
  activity" design involves problem solving

53
My view on the classical SIP approach to design
elements of critique

• SIP defends a too systematic view of design
• ? impoverished view of design that does not
  render the design activity as actually
  implemented by designers on "real", industrial
  design projects

54
My view on the situativity approach to design
elements of critique

• allusive character with respect to theoretical
  paradigms
• methods used for data collection analysis and
  for modelling offer no tools for deriving
  higher-level descriptions from the data
• data has an anecdotal flavour ("Interesting, but
  so what?")results are unlikely to be replicable,
  conclusions unlikely to generalise across
  situations
• extremely rich descriptions of "unique"
  designers implementing "unique" activities in
  "unique" situations

55
My view on the situativity approach to design
elements of recognition

• attention for
• situations, i.e. combinations of designer and
  task
• construction of representations (cf. "problem
  setting" and " problem (re)framing", "active" and
  "constructive" aspects of the activity)
• ever evolving representation of problem, I.e.
  state of design project, of to-be-designed
  artefact

56
Problem-representation construction, solution
generation and solution evaluation details (3
slides)
57
Problem-representation construction, solution
generation and solution evaluation

• solution generation may take different forms
• from the simple "evocation" of solutions existing
  in memory
• to the "elaboration" of new solutions out of
  mnesic entities without any clear link to the
  current problem

58
Problem-representation construction, solution
generation and solution evaluation

• evaluation two types corresponding to two
  functions
• solution evaluation at the action-execution
  level of the design activity (classical function)
  
• design action evaluation at the
  action-management level
• select a design action proposal
• guide the design process
• solution evaluation
• confront solutions to evaluative references
• ? selection of a solution proposal

59
Problem-representation construction, solution
generation and solution evaluation

• evaluative references
• prescribed, constructed and deducted evaluation
  constraints
• designers differ with respect to their capacity
  to bring evaluative references to their task
  (Bonnardel)
• evaluation modes
• depend on the evaluative reference analytical,
  analogical and comparative evaluation modes

60
Design strategies details (13 slides)
61
Decomposition and planning in design

• decomposition
• method advocated by design methodologies
• "fundamental strategy in design" ? conclusion of
  early empirical cognitive design studies (esp. in
  software design)
• several hypotheses concerning this conclusion
• confusion between structure of the actual
  activity and structure of its result
• participants were "novice" designers (students)
  working on artificially restricted, small
  problems
• by definition, design problems are decomposed,
  i.e. transformed into several other problems to
  be solved
• but decomposition can take different forms

62
Decomposition and planning in design

• important function of decomposition planning the
  activity
• empirical cognitive design studies on
  professional designers working on industrial
  design projects
• no systematic implementation of pre-established
  structure (hierarchical or other structured
  problem decomposition)
• no systematic implementation of plan in actual
  solution development
• the observed "scattered" way of designers'
  proceeding is not due to their nonchalance or
  incompetency (cf. factors underlying
  opportunism)

63
Planning and organising design the opportunistic
organisation of design

• plan for an activity ? actual organisation of the
  activity
• design activity is "opportunistically" organised
• global model based on a blackboard approach
  (Hayes-Roth and Hayes-Roth)
• distinguish two levels action execution and
  action-management (control)
• (i) call for action proposals(ii) proposal of
  one or more actions(iii) evaluation of the
  proposed action(s), leading to selection of
  one action (iv) execution of the selected
  action(v) back to (i)

64
Planning and organising design the opportunistic
organisation of design

• action selection
• evaluation criterion of cognitive economy
• action proposal
• opportunities must be perceived
• perception is (also) data-driven
• six types of categories of data as factors
  leading to the opportunistic organisation of
  design (Visser)

65
Planning and organising design the opportunistic
organisation of design

• Example of data category. Representations used
  for the current design action activate mental
  representations of design objects because of
  evocation-guiding relationships existing between
  the two sets of representations.
• Example (Functional specification study, Visser)
• designer's plan specify 1st-phase O1 before
  2nd-phase O2 tooling
• designer's actual activity completely
  intertwining of both specifications designer
  "takes advantage" of 1st-phase O1 specification
  in order to specify O1's "corresponding" O2
  operation (by adapting O1's specification)
• interpretation analogy between 1st-phase and
  2nd-phase operations
• other consequence O2 specification "makes" the
  designer "think" of omissions and errors made on
  corresponding O1 operations

66
Planning and organising design the opportunistic
organisation of design

• opportunistic organising one's activity
• is not restricted to inexperienced designers to
  the contrary
• is not translating deteriorated behaviour when
  confronted with particularly difficult tasks
• analysis of 15 empirical cognitive design studies
  (Visser, )
• even if designers possess a pre-existing solution
  plan for a design problem
• and if they can and, in fact, do retrieve this
  plan to solve their problem (which is often
  possible for expert designers during "routine
  design")
• yet if they also perceive other possibilities for
  action ("opportunities") (which is often the
  case in real design)
• and if they evaluate the cognitive and other
  cost of possible actions, (as they will do in
  real design)
• the action selected for execution will often be
  an opportunity that is chosen because of its
  cost, I.e. another action than that proposed by
  the plan

67
Planning and organising design the opportunistic
organisation of design

• expert designers possses structured plans and
  they use them
• BUT they also use other resources to select
  successive design actions
• conclusion pre-existing plans
• are only one of the various action-proposing
  knowledge structures used by designers
• may be interesting from a cognitive-economy
  viewpoint because of their schema nature
  variables with default values
• BUT if other knowledge structures propose
  relatively more economical actions, designers may
  deviate from such plans

68
Planning and organising design the opportunistic
organisation of design

• differences between results from laboratory
  experiments and from field studies
• explanation
• only in "real" design, designers
• perceive several possibilities for action
• take into account the cost of actions

69
Decomposition and planning Planning and
organisation

• conclusion
• designers' problem decomposition leading to an
  action plan, does not mean that the designers are
  going to observe this plan
• the actual organisation of design activity will
  generally be opportunistic, following only in
  part the plans that designers have developed!

70
Decomposition and planning Planning and
organisation

• characteristic of an opportunistically organised
  activity
• no independent stages in design problem solving
• design problem solving is not a process going
• from analysing "the" problem specifications
• to synthesising "the" solution
• problems do not pre-exist to solutions
• both are built up and elaborated simultaneously
• the intermingled character of "analysis",
  "synthesis" and "evaluation"

71
Reuse vs. design from scratch

• "reuse" use of knowledge that is at the same
  abstraction level as the target for whose
  processing the knowledge in question is being
  retrieved
• empirical design studies
• domain of software design, esp. object-oriented
  software
• global model of reuse in problem-solving0. const
  ruction of a representation of the target problem
  1. retrieval of one or more sources
  2. adaptation of the source into a
  target-solution proposal 3. evaluation of the
  target-solution proposal4. integration into
  memory of the resulting modifications in
  problem and solution representations

72
Reuse vs. design from scratch

• reuse rather than design from scratch
  when?0. construction of a representation of the
  target problem decision reuse rather than
  design from scratch 1. retrieval of one or more
  sources
• cost of reuse (cf. cognitive economy underlying
  design organisation)
• conclusion of experimental studies of analogical
  reasoning source retrieval occurs seldom
  spontaneously

73
Simulation

• important evaluation strategy
• but also a generation strategy simulation for
  knowledge access
• mental simulation
• physical simulation (mechanical and industrial
  engineering design)