Mention ambiguity to a psychologist and the first thing that will come to mind will be one of the famous ambiguous figures popularised by the Gestalt psychologist in the early 20th century, perhaps Boring’s hag, Dunker’s duck-rabbit, or the geometrically tantalising Necker cube.
They are fun to look at because our perception of them changes, apparently spontaneously: one moment we are looking at an elegant but shy young woman, the next a grinning old witch. They are more than intriguing puzzles, though, because they demonstrate that the percepts we form in our mind are based on more than the sensory information we can gather from the world. Perception is a great deal more than sensation. We actively construct our perception of the world from sensory information, and when two equally plausible alternatives are possible our mind has to choose one.
You might think that ambiguous figures are trivial exceptions to the normal operation of our perceptions, which present us with a truthful internal model of the world around us, but they actually show how perception normally works. The sensory information we obtain form the world is inherently noisy, indeed ambiguous, and our minds perform incredible feats of interpretation to present us with a stable, comprehensible perception. Just think about how difficult it is to derive any perception from the noisy pressure waves hitting your two ears, unpredictably out of synchronisation, and the patterns of activity on the rods and cones of your retinas, after light has filtered through the cornea, vitreous humor and blood supply. It makes me want to lie down in a quiet, dark room.
The world is very ambiguous – but our minds resolve that ambiguity and choose a single answer, most of the time. Almost all of the time. These classic images are tantalising because their equipotent resolutions are so finely balanced yet incongruent, imbued with different meanings to us, that when our percept changes, we feel the world shift.
Ambiguity in this sense is at the heart of psychology. Many experiments put people in situations that are ambiguous, where a person could perceive different things or choose alternative actions; and the psychologist counts their choices, or times their decisions. Sometimes one outcome is counted as ‘correct’ and others as ‘error’, and we contrast error rates, but often we are just interested in the choices people make, and whether or not changes in the situation affect the proportions or rates of those choices in accordance with our theoretical predictions. If everyone behaved identically in a given situation, like particles, waves, chemicals and reagents seem to, psychology wouldn’t be as fun to do, nor as difficult. Psychology is about understanding the ambiguity of human behaviour, and predicting it.
But I also have another view of ambiguity. As a psychologist I have worked in the field of human factors, helping designers build technologies for people to use to accomplish a variety of tasks. In this world, ambiguity is something to be avoided, because a device interface that is ambiguous is hard to learn, leads to error and frustration, and low usability. And that means low sales. Most of the human-computer interaction literature is about removing ambiguity, making interfaces ‘intuitive’ and clear; informative and guessable; transparent. There should be an obvious thing to do at any moment, every object should ‘afford’ an action, and every feature of the device should be apparent. The user’s mental model of the device should be consistent with its function and operation.
Imagine buying the latest device, opening the box and taking out a sleek black slab with some silver pieces of metal shining on the side. What does it do? Do you press the bits of metal, or slide them? Which way? Which way up do you hold it? Do you hold it at all? This hyper-ambiguous device can only be used if you bring knowledge to bear upon it. You need to be instructed, or have expectations about it, that resolve the ambiguity.
If you think about what makes you frustrated with technology, it will come down to ambiguous situations that offer alternative actions, but where only one of them is correct – and unlike the psychology experimenter’s normative use of error, an interaction error will result in failure, and even an error message. How hard can it be to put a USB stick into an unfamiliar computer? An error message has appeared – should I click ‘Close’ or ‘Cancel’? Or is it just ‘OK’? These are the situations in design where ambiguity is to be avoided, and where design should guide the user clearly through the interaction without requiring them to exert cognitive effort to consciously disambiguate the situation. You can see the same needs in safety critical systems, where ambiguous control panels have been shown to lead to disasters in nuclear power stations and aircraft. When the operator or pilot chooses the wrong action, this is not human error, it is design error, and too often it is due to ambiguity. In this part of my psychology, ambiguity is something to be minimised and avoided.
Now imagine your new graphics editor. You open the application to draw something, and it is totally unambiguous about what to do next, because all you can do is choose a shape, and then all you can do is choose its colour, in fact, you have to step through a series of choices until you have defined all of the shape’s possible attributes and nothing is left undefined or ambiguous. You cannot make an interaction mistake. This graphics editor is completely unambiguous at every step, and is therefore totally useless for drawing, because it limits the options available to you, and by forcing you to define everything, leaves nothing open to accident. It minimises spontaneity and stifles creativity, because what you get is what you define, and not what you spot happening by chance, through unintentioned coincidence.
To support creativity in this way, it isn’t the interface that needs to be ambiguous, so much as the interaction. The ambiguity lies in the user’s intentions, and the device should offer options rather than ambiguity. Redesigning the interface to allow the user to be ambiguous about their commands, and for the application to interpret them, and there is scope for chance to produce something that the user then perceives as useful or valuable or aesthetically pleasing.
Should a device like this be ambiguous in behaving in different or unpredictable ways? That might make it more likely to help the user spot novel or unintentioned products, and so support found creativity, but it will also be frustrating as the user tries and fails to reproduce the effect they want. It all depends upon the user’s goal, and whether they have something definite in mind, or whether they will accept one of a range of outcomes. Being able to identify this tolerance of ambiguity in the user’s task would help a tool know when to be ambiguous and remain usable, supporting creative discovery.