Lectures: Visual Perception 6

Contexts and Expectations

Someone once said that there is no meaning without context. Various kinds of context are important in shaping our interpretation of what we see. As a reminder of the importance of making clear what is meant by the importance of 'context' in perception I briefly list here several very different uses of the term. However, I would not suggest that in practice tidy distinctions can always be usefully made.

The largest frame is that of the historical context of perception. Some theorists, such as Marshall McLuhan (1962), Walter Ong (1967) and Donald Lowe (1982), have argued that there have been shifts over time in the human 'sensorium' - that is, in the 'balance' of our senses or the priority which we give to some compared with others. Such argue that in western urban cultures we have come to rely more on sight than on any other sense (this was referred to in Visual Perception 1 as 'ocularcentrism').

Another major framework is that of the socio-cultural context of perception. Just as there may be subtle differences in human perception over time there may also be differences attributable to culture. Some of these were alluded to in Visual Perception 3. Constance Classen (1993) in her book Worlds of Sense shows that different cultures accord priority to different senses - the Ongee of the Andaman Islands, for instance, live in a world ordered by smell.

A native American Indian writer called Jamake Highwater, who is of Blackfeet/Cherokee heritage, draws attention in the following extract to radically different ways of seeing the world:

Both the historical and socio-cultural context of perception are vast themes which will not be explored further here, but such studies do help to emphasize that 'the world' is not simply indisputably 'out there' but is to some extent constructed in the process of perception. Within a given socio-cultural context, there are widely-shared interpretive conventions and practices. Whilst the basic processes of human perception are largely universal there is scope for subtle but significant variations over space and time.

Several other kinds of context are commonly referred to. I have referred already, in Visual Perception 3, to the importance of individual factors which can have an influence on perception. An emphasis on the individual as a context emphasizes the role of the various long-term characteristics of individual perceivers such as values, attitudes, habits and so on. An emphasis on the situational context considers such transitory situational factors as goals, intentions, situational constraints and contextual expectancies. Finally, an emphasis on the structural context stresses structural features and relationships (such as the relationship between one line and another) 'in' what is perceived - though the extent to which there is agreement about even such low-level formal features may vary.

Five main definitions of the scope of the term 'context' have been listed here in relation to their potential influence on perception:

Whilst it may be useful to be alert to the very different meanings that the word 'context' can have, disentangling them is problematic.

A very well-known study by Bugelski and Alampay (1961) can be seen as showing the importance of situational context. Their experiment is often used as an example of the influence of what psychologists call 'perceptual set': a predisposition to perceive something in relation to prior perceptual experiences. Perceptual set is broader than situational context, since it may involve either long-term (for instance, cultural) prior experience or, as in this case, short-term or situational factors (Murch 1973, 300-301). Groups of observers in the experiment were shown an ambiguous line drawing which was designed to be open to interpretation either as a rat or as a bald man wearing spectacles. Prior to seeing this image, two groups were shown from one to four drawings in a similar style. One group was shown drawings of various animals and the second group was shown drawings of human faces (see illustration below). A control group was shown no pictures beforehand. 81% of the control group reported seeing the ambiguous image as a man rather than a rat. The more pictures of animals that the 'animal' group had seen, the more likely they were to see a rat rather than a man (with 4 prior images of animals 100% then saw a rat). From 73-80% of the 'faces' group subsequently saw a man rather than a rat.

The influence of perceptual set has also been explored in relation to the famous image shown below:

This image was designed to be interpreted as either a young woman or an old woman. It was introduced into the psychological literature by Edwin G Boring (1930) (though it was published by the British cartoonist W E Hill in 1915, and is thought to be based on a French version of 15 years earlier). It is sometimes given the chauvinistic label of 'The Wife and the Mother-in-Law'. In order to study the role of perceptual set Robert Leeper (1935) had the image redrawn in two 'biased' forms: one which emphasized the old woman and the other which emphasized the young woman (see image below).

Leeper varied the conditions of viewing for five groups. A control group was shown only the ambiguous drawing, and 65% of this group spontaneously described the image as that of a young woman. The second and third groups were first given a verbal description of the old woman and the young woman respectively. The fourth and fifth groups were first shown the 'old' version and the 'young' version respectively. Groups 2 to 5 were then shown the original ambiguous image. Leeper found that each of the primed groups was 'locked-in' to their previous interpretation. 100% of group 5, which had seen the young version first, interpreted the ambiguous image as a young woman. 94% of group 4, which had seen the old version first, reported seeing the old woman in the ambiguous image. The percentages opting for each interpretation amongst those given verbal descriptions were much the same as for the control group. Gerald Murch (1973, 305) was unable to replicate these findings (94% of his control group first saw the young woman) and suggested that the image was by then so well-known that this may have influenced the results.

Particular situational contexts set up expectations in the observer. Bruner and Postman (1949) conducted an experiment in which playing-cards were used, some of which had the colour changed from red to black or vice versa. The cards were exposed in succession for a very short time. Subjects identified them as follows:

Interpretation here was dominated by what the situational context suggested that people ought to be seeing. A shorter time of exposure was necessary for people to name the normal cards than the anomalous ones.

In one experiment, Steven Palmer (1975) first presented a situational context such as a kitchen scene and then briefly flashed on a target image. When asked to identify a loaf-like image, people who had first seen the kitchen correctly identified it as a loaf 80% of the time. Obviously, a loaf of bread is the kind of thing you’d expect to find in a kitchen. They were asked to identify an image like an open US mailbox and an image resembling a drum - two objects not usually associated with the kitchen. The images were a little ambiguous: the mailbox was a little like the shape of a loaf with a slice of bread lying next to it, and the drum could have been interpreted as the lid of a jar. People who had first seen the kitchen only identified these as a mailbox and a drum 40% of the time. The ability to identify objects was affected by people’s expectations concerning what is likely to be found in a kitchen.

I have mentioned that situational contexts generate certain (short-term) expectations but it is worth noting in passing that expectations may also be set up by longer-term influences - such as by stereotypes, prejudices and past experience.

To return to contexts, here is an example of structural context. This pattern of circles is known as the Ebbinghaus (or Titchener) illusion. It is an illusion of relative size (or more strictly, area). Here the formal relationship between the parts of the image leads the small white circle (which is the same size in both images) to seem larger in the structural context of the tiny black circles than amongst the large black circles. There is no shortage of examples of the role of structural context amongst the geometrical illusions which can be found in psychology textbooks so no further examples of the role of structural context will be discussed here.

At this point it is useful to introduce schema theory briefly. A schema (plural 'schemata' or 'schemas') is a kind of mental template or framework which we use to make sense of things. Particular circumstances seem to activate appropriate schemata, which set up various standard expectations about such contexts. Such schemata develop from experience. They help us to ‘go beyond the information given’ (as Jerome Bruner famously put it) by making assumptions about what is usual in similar contexts. They allow us, for instance, to make inferences about things which are not currently directly visible. The application of schemata and the expectations which they set up represents 'top-down' processes in perception (whilst the activation of schemata by sensory data is a 'bottom-up' process). A good example of the role of top-down processes is where you think that you recognize someone in the street and then realize (from sensory data) that you are wrong. We are often misled in this way by situational contexts, by wishful thinking and so on, ignoring contradictory sensory data in favour of our expectations.

In an experiment by Brewer and Treyens (1981), individual participants were asked to wait in an office. The experimenter said that this was his office and that they should wait there whilst he checked the laboratory to see if the previous participant had finished. After 35 seconds, he returned and took the participant to another room where they were asked to recall everything in the room in which they had been waiting. People showed a strong tendency to recall objects consistent with the typical ‘office schema’. Nearly everyone remembered the desk and the chair next to it. Only eight out of the 30 recalled the skull (!), few recalled the wine bottle or the coffee pot, and only one recalled the picnic basket. Some recalled items that had not been there at all: 9 remembered books. This shows how people may introduce new items consistent with the schema.

In an experiment by Baggett (1975) participants were shown a series of simple line drawings telling a story. One story showed a long-haired man entering a barbershop, then sitting in the barber’s chair, and finally leaving the shop with shorter hair. In a later test they also saw a picture showing the actual haircut, which had not been present originally. People were fairly good at remembering that this picture had not been present if the test followed immediately after the initial showing. However, if the test occurred a week after the initial presentation most people claimed that they had seen the haircutting picture in the original sequence. This shows the way in which we incorporate in our memories inferences derived from our schemata. This experiment was concerned with memory rather than perception, but it is difficult to separate these processes if you take the stance that no perception is 'immediate'.

home :: admin :: lectures :: reading :: assignments :: modules :: tutor