These lecture notes were not written as a course handout, but as a resource for lectures. Therefore, references and comments will not always be complete.
The role of language in human factors is a very rich area -- especially once one considers the human factors in a wider context, that is, in instructional / communicative / written material and also the problems and potentials for natural language communication with machines.
As with visual perception, our interpretation is based as much on what we expect as on what we actual see or hear. "No head injury is too trivial to be ignored." (cf. "No flowers are too expensive to be bought.")
Key principles:
For more on the proper use of language in human factors, see Strunk and White.
see Oborne pp. 73-100.
Increasingly computer systems are designed to help people solve problems, make decisions, and other complex tasks. This is a classic area where the formalist approach leads one astray. It is tempting to assume that the brain must have some form of logic inside it, that it must function 'logically'.
I think that the humans behave logically is likely to be a fairly good hypothesis, but that logic is not based on the task level. Logically in this case means that the knowledge the user has, whatever that knowledge may be, and as incorrect and as incomplete as that knowledge may be, will be applied in a consistent, logical manner. If the user behave differently in the same situation, it is the situation that is the same to the observer, not the user. And if the behaviour is not optimal, that again is in the eye of the observer; the user is quite likely to think that they are doing the best they can.
Also, in emergency situations -- when functioning at Rasmussen's "knowledge-level" -- the knowledge has to be used in a problem-solving manner. Fault diagnosis is a classic example of this.
Many decision-support systems aim to try and support the process of formalising the information contributing to the solution / decision.
However, human problem-solving is not based on logic or even rationality, in terms of using the information to the full. This makes good decision support systems very necessary, but very difficult to design.
Formal analyses of problems do not give accurate predictions of difficulty.
Plausibility of conclusion is far more influential than validity of logic.
People do not use even simple logical rules: if p then q therefore if not-q then not-p and will even deny their relevance or validity. This is not to imply that their internal reasoning is not logical, just that they use rules that look more like, if it's blue based on my experience, then propose blue.
We do not take account of a lot of potentially relevant information when making decisions.
We seem to treat negative instances, responses, experiences as containing no information - and, thus, we do not seek them out -- a critical error in fault diagnosis tasks (e.g. software testing).
We often misuse evidence involving numbers -- either believing too much in large numbers (it always happens), or being influenced by small numbers (1 in a million happens quite often). And certainly rarely appreciating the difference.
Influenced by "the number you first think of" -- even when known to be a random number. Which has a greater population, Ireland or Indonesia? (at least in studies done in America).
Decision-making and problem-solving seem to be thoroughly rooted in our world knowledge and world experiences.
We seem to imagine a world in which the propositions are true and read off the solution. Information that cannot be easily imagined (negatives, sample sizes) does not easily contribute to the decision.
In trying to understand unfamiliar machines we try to construct a mental model (imagined world) in which we understand the device. This model enables us to operate the device without recourse to much memorising. The correspondence between the imagined world and the real world is not important, as long as the operations make sense in the imagined world.
For example, Don Norman's fridge operation:
Consider how one might build a mental model of this fridge. What other devices is it similar to? How does it work? What components does it have?
Metaphors for design exploit this process by prompting an appropriate mental model of the device - need to be careful though. For example, <tab>, <cr> in word-processing confuse very large numbers of skilled typists, since <tab> and <cur> on a typewriter are quite different (they are not characters, they are more like formatting instructions).
Individuals in groups do not behave as they would on their own. See a social psychology text book for numerous examples, including:
These effects are more important when considering the design and engineering process, rather than the process of actual use, but software to support group activities might need to be designed to prevent such group influences.
There has been observation of the fact that groups seem to make riskier decisions than individuals -- sometimes even than any of the individuals would make alone.
However, there are also situations where groups can be more conservative than individuals.
Nevertheless, one should not expect groups to reach same decisions as individuals -- predicting whether groups will be better or worse may be much harder.
Attributions are how we explain our own actions and the actions of others. There are some key regularities in the way we do this:
Cognitive dissonance is term used to describe effect of perceiving increased value in things we've done without high reward - in order to explain our having done it we increase our perceived value of it.
Computer use, with its high failure rate (i.e. plenty of negative experience) is a prime arena for attributions. One explanation of why some people take to it so well is an attributional style difference, rooted in expectations and confidence. Some people seem to successfully take credit for success and blame the machine for failure, whilst other computer users seem to get bogged down in a cycle of blaming themselves for failure and crediting serendipity for success. Designing systems that help people make the right attributions (error messages usually imply it is your fault, not the machines!) is a challenging area.
Physical and psychological health are both a complex blend of physical and psychological factors -- a key one appears to be the beneficial effect of being in control of aspects of one's life. For example, problems such as Repetitive Strain Injury (RSI) do not tend to occur in people who choose to use computers (hackers working hours and hours per day). RSI seems to be limited to non-discretionary users. Rest breaks are necessary, but length / frequency are less important than being in control of when to take a rest break.
Most human factors work with personality has dealt with accidents. Many psychologists now doubt the validity of these personality tests.
Extroverts seem to have more accidents than introverts.
Does not seem to be a factor (except in so far as it may be extroversion).
There is some indication that some people have a higher tendency to take risks, who require a higher level of risk to get the same feeling of success. Seems to relate to accident numbers.
But is this tautologous? Does it relate to accident causation? Or is it simply a correlation? That is, accidents happen in risky environments, the more time you spend in such environment the more likely you are to have an accident - even if you are no part of the cause.
Communication structures are key to the functioning of organisations. The communication networks can be studied and evaluated, as chains, nets, spokes, hierarchies, and so on. The effectiveness of each of these varies -- in part in relation to the tasks being performed. However, many new technologies do not support the existing structures -- indeed, many undermine any structure at all.
Organisations are full of procedures, which describe the way things should be done. These are invariably different from reality. Many organisations have suffered disastrous computerisation because they computerised official procedures without recognising that the reality was different. Often, the operators are not aware of the differences, and even when they are they may be reluctant to tell a consultant who will be reporting back to management. The 'breach' of procedures may only be the fact that procedures written as though they are performed by one person are, in fact, shared amongst a number -- or they may be written down as though they occur at one time without interruption when, in business fact, that procedure is always interleaved with numerous other similar tasks and problem solving.
Grice, H. P. (1975). Logic and conversation. In P. Cole & J. L. Morgan (Eds.), Syntax and semantics III: Speech acts. NY, NY: Academic Press.
Strunk, W., & White, E. B. (1979). The elements of style. NY, NY: Macmillan. Cheap, short, to the point.