Human-computer interaction Chapter 3

Premises => Algorithms => Conclusions. – Example : ... The conclusion always derives from the premises by logical operations but this does not ... Another application are the « expert systems » : in order to choose the ... Research state : information space containing all data and procedures that will be used to solve the ...
264KB taille 6 téléchargements 250 vues
Human-computer interaction Chapter 3 – Reasoning and problem solving

Interaction homme-ordinateur

Page 1/III

Petre Dimo - 2008

Definitions ●



Reasoning –

The process by which we use the knowledge we have to draw conclusions or infer something new about the subject of interest



Types of reasoning ●

Deductive



Inductive



Abductive

Problem solving –

Use information we have to find solutions in new situations ●

Gestalt (or form) theory



The problem space (Newell et Simon)



Analogy

Interaction homme-ordinateur

Page 2/III

Petre Dimo - 2008

Deductive reasoning ●

Premises => Algorithms => Conclusions –

Example :

We don't work on weekends. We are saturday Saturday is the first day of the weekend Conclusion : I do not work today –

The conclusion always derives from the premises by logical operations but this does not mean that the premises were correct !



Deductive reasoning may lead to false conclusions

Certain people are babies Certain babies cry conclusion : certain people cry ??? FALSE, since it is not said that ALL babies are people... those who cry may not be people Interaction homme-ordinateur

Page 3/III

Petre Dimo - 2008

Inductive reasoning (1) ●



Inductive reasoning corresponds to the generalization that we make from cases we know to infer information about case we don't know. The facts are true (experience) but the generalizations (making a law) may not be : one counterexample is enough to disprove the « law ». Example 1 : –

The first line hereafter shows 5 hexagones ; it is asked to add the 6-th one, selected between those shown in the second line.

Interaction homme-ordinateur

Page 4/III

Petre Dimo - 2008

Inductive reasoning (2)

Interaction homme-ordinateur

Page 5/III

Petre Dimo - 2008

Inductive reasoning (2) ●



Inductive reasoning helps us learn and orient in a changing environment. But the law established by inductive reasoning must be constantly verified Example 2 : Wason cards

7 E 4 K Each card has a number on one side and a letter on the other side. Find which cards must be turned on to prouve the following « law » that was « established » by previous tests, not presented here : If a card has a vowell on one side, it has an even number on the other side What happens if we omit saying that each card has a vowell on one side and a number on the other ?

Interaction homme-ordinateur

Page 6/III

Petre Dimo - 2008

Abductive reasoning ●

Abduction : it is the type of reasoning used to determine the causes of a given effect : –

The result is observed



One knows causes to effect possible relations



A hypothesis to explain the causes is put forward



The hypothesis is checked

This tipe of reasoning is common to police inquieries and to medical diagnostic, but it is also used in everyday life. Another application are the « expert systems » : in order to choose the most appropriate hypothesis, one sets up a mechanism to evaluate the « value » of each one proposed.

Interaction homme-ordinateur

Page 7/III

Petre Dimo - 2008

Problem solving : gestalt theory ●





Gestalten in german means « to structure, to give form » ; it has no equivalent in other languages and for this reason the word is used « as is » in french, english, romanian... The main ideas are : –

We are pre-programmed to percieve forms that we use constantly to understand our environment, by grouping and restructuring them to solve the new problems



The « whole » is not the sum of the parts : we percieve the whole before percieving the details

This principles are the basis for the « gestalt » theory and are also useful guides to HCI design. The theory establishes the following « laws » :

Interaction homme-ordinateur

Page 8/III

Petre Dimo - 2008

The « laws » of the gestalt theory ●

The good form : an informal set is first percieved as a simple form, stable and symmetric



Continuity : close points are first percieved as forming a line



Similitude : in a set we try to assemble similar objects





Common destiny : the objects moving simultanously in the same direction are percieved as being part of the same form Closure : it is easier to percieve closed forms surrounding a surface

Conclusion : we combine our perception with our knowledge and thus we add information which is absent in the stimuli

Interaction homme-ordinateur

Page 9/III

Petre Dimo - 2008

Gestaltic perception ●

Example :

Although attractive, this theory leaves unexplained many aspects of the problem solving mechanisms. Interaction homme-ordinateur

Page 10/III

Petre Dimo - 2008

Problem solving : the problem space

Initial state Search

Desired state

Search Problem space (Newell, Simon - 1972) Problem space : ● Initial state : problem statement and known conditions ● Research state : information space containing all data and procedures that will be used to solve the problem ● Goal state : target description Limitations : ● The model works if both the initial and final states are well defined and well known. Sometimes this is impossible and the state definitions become a part of the problem, like in the case of functional specificatons in programming. Interaction homme-ordinateur

Page 11/III

Petre Dimo - 2008

Problem solving exercise Identify the goals and operators involved in the problem 'delete the second paragraph of the document' on a word processor. Now use a word processor to delete a paragraph and note your actions, goals and subgoals. How well did they match your earlier description?

Interaction homme-ordinateur

Page 12/III

Petre Dimo - 2008

Problem solving : analogy





The analogy is used to solve a new problem by looking for similitudes with known situations and events, for which the problem was solved. Example : a doctor has to blast a malignian tumour with very strong rays. But he knows that the surrounding healthy tissue will also be destroyed ; however, if he lessens the rays' intensity the healthy tissue will be preserved but the tumour will remain. How should the doctor proceed ?

Interaction homme-ordinateur

Page 13/III

Petre Dimo - 2008

Problem solving : the following story can help ?

A greek general attacking an ennemy city had to avoid the main road, since the defenders could easily stop his soldiers. He decided to split his army in small groups which attacked arriving on several roads simultaneously. The manoeuvre was successfull.

Interaction homme-ordinateur

Page 14/III

Petre Dimo - 2008

Skill acquisition Understanding the skill acquisition mechanisms became more and more important with the advent of computer and it's use in computer assisted training. The help systems integrated to computer applications are also addressing skill acquisition mechanisms. ●

Concept acquisition –



Skill acquisition –



Repetition's role

Experts vs novices behavour (chess ?)

Methods –

Structured acquisition



Try and error

Interaction homme-ordinateur

Page 15/III

Petre Dimo - 2008

Human errors Understanding human errors' reasons is important for decision taking processes. Some errors may have catastrophic effects. There are many reasons why humans make errors ; wrong man / machine interfaces can be a cause (the case of the Spitfire pilots) ●

Bad design



Response time



Incompetence



Lack of attention or concentration –





To many informations arriving simultaneously can divert the user

Context change –

Unexpected (example : a difference between the real behaviour of a device and it's description in the documentation or explanation during training)



Loose preparation (example : a behavioural change is asked but the operator was not informed or incomplete information was provided).

Natural or provoked emotion

Interaction homme-ordinateur

Page 16/III

Petre Dimo - 2008