L3 Flashcards
Point of understanding goal oriented interaction
the ways that we can use cognitive theories of planning, learning and understanding to understand user behaviour, and what they find hard
This approach is founded on a meta-theory of first-wave HCI, that “user interaction can be modelled as search”.
General purpose search algorithms are familiar in computer science, where an objective function can be combined with a state space and dependency graph to recursively search for optimal solutions using a wide variety of breadth-first or depth-first strategies for reducing the distance to some goal or target.
What needs to be specified to model interaction as goal directed search. What is an evaluation method which proceeds in this way?
User goal definition, model interaction as a search process which gets user closer to goal
Cognitive walkthrough
What is cognitive walkthrough
analyse the user interface by identifying the next user goal,
determining whether the necessary actions are available,
ensuring that they are labelled in a way that the user will recognise them,
and confirming that the system will give appropriate feedback of progression toward the goal
Downsides of modelling HCI as goal directed search
many situations in which users may not be able to achieve an optimal goal,
and where it may not be possible for designers to correctly anticipate what the user’s goal is.
The above process works well for simple user interfaces, in which there is an exact specification of what the user ought to do,
and where their understanding of their goals is consistent with the designer’s view.
Describe how interaction as goal oriented search fails in more complex situations
In more complex situations, even where an optimal solution exists, the amount of search time necessary to find it may be too large to be justified.
Models of goal-directed planning that take the expense of computation into account are described as bounded rationality. Rather than optimising strategies, users often engage in satisficing strategies, where they follow a plan that is satisfactory, rather than optimal, within constraints.
What is prospect theory
prospect theory describes human behaviour in terms of a utility model that considers the outcome of possible actions, with weighting of estimated benefits by likelihood. Where optimising search assumes complete knowledge of the state space, prospect theory assumes only that people choose actions based on estimated utility.
Name six heuristics/biases
Availability heuristic Affect heuristic Representativeness heuristic Loss aversion Expectation bias Bandwagon effect
What is the availability heuristic
availability heuristic, in which reasoning is based on examples easily to hand
Affect heuristic
- affect heuristic, basing decisions on emotion rather than calculating cost and benefit,
Representativeness heuristic
in which probability is judged based on resemblance to a class of similar situations
Why do humans apply biases?
Humans also apply biases to ensure that the consequences of estimation error are within tolerable bounds
Loss aversion
that losses hurt more than gains feel good
Expectation bias
people observe results they expected
Bandwagon effect
People prefer to copy actions taken by other people
Security implications – all users could be using your system incorrectly
What is the attention investment theory of abstraction use?
The Attention Investment theory of abstraction use is a model of end-user programming, which explains why users without prior experience of programming may take decisions that favour repeated manual actions rather than automated shortcuts
This occurs where automation involves forming an abstract specification, such as defining a regular expression for search and replace
The benefit of automation is saving time and concentration in future, but abstract specification (programming) takes time and concentration.
There is also the risk of “bugs” that might result in the automated solution going wrong, and perhaps resulting more manual effort to fix up the consequences.
So the utility function compares future saving of attention from programming vs costs of concentrating on a risky strategy.
Biases such as loss aversion are likely to apply, and bounded rationality will apply, since deciding what to do takes even more concentration.
What does behavioural economics tell us about much routine computer use?
In the extreme case of minimal concentration, it is worth remembering that much routine computer usage is carried out simply on the basis of memorised patterns of interaction (for example, repeatedly pushing the clear button on a calculator, the ‘walk’ button at a traffic light etc), with no clear mental model or goal underlying these.
Describe the limitations of goal based HCI
Rational models of planning assume that the user doesn’t make mistakes, which is unrealistic even for experienced users
If we wanted to account and anticipate all user actions, including errors, we would need a cognitive model of why the error occurred - that is, a decision process that is not consistent with the identified goals, constraints, and search space
This might include information loss due to cognitive limitations, incorrect mental models, or misleading designs.
Anticipating all of these factors would need description of a user journey that accounts for problem identification, diagnosis, debugging, testing, iteration, and many other procedures that characterise the user’s own activity as a kind of design process.
What is persuasive design
Consider how users might choose alternative goals, or modify their goals
These methods are useful in applications such as software systems to reduce energy consumption, promote … apply nudge methods to modify the biases that underlie inappropriate goals
Might be recognised by users who may respond negatively to a design that is paternalistic
Characteristics of wicked problems
Cannot be addressed with goal-based problem solving methods
Social problems
10 characteristics: formulation, stopping rule, good/bad, no ultimate solution test, one shot, not enumerable solution set, unique, symptom, choice of explanation determines resolution, planner has no right to be wrong.
