Lecture 1: Sex age and ethnicity Flashcards
why have faces evolved?
We first must understand what faces are for and how they may have evolved, since this places important constraints on what messages can be signalled by faces and how these can be perceived. Face plays important role in biological functions- eyes and ears are spaced to allow us to perceive distance.
what do structural differences lead to?
General and species specific factors mean that all human faces are remarkably similar in basic form. Despite this, there are subtle differences that make every face unique so that faces play an important role in the identification of individual members of our highly social species, and
intro to social categories
Our faces convey a lot of information about our social identities.
From looking at someone’s face we can infer some details of their racial background and decide whether they seem old or young, male or female, attractive or unattractive, friendly or unapproachable, intelligent or unintelligent, and so on.
Some of these inferences (such as age or sex) are remarkably accurate.
talk about stereotyping
The standard view in psychology, derived from classic theorising of Lippmann (1922) and Allport (1954), is that stereotyping is useful to the person doing the stereotyping.
An obvious benefit is that it creates cognitive shortcuts based on past experience, but it also has more social functions in establishing social identities or justifying the status quo.
Noticing that someone is a man or woman, or young or old, establishes a set of expectations that can save some of the mental effort of dealing with everyone you meet as a truly unique individual. The downside, of course, is prejudice, where you don’t look beyond the stereotype.
experiment on how we seem to acquire facial judgments in a non-conscious manner
Lewicki 1986- experiment using photos of females with long or short hair and read brief vignette of description, so that three faces with long hair were shown accompanied by descriptions emphasising their kindness, and three faces with short hair accompanied by descriptions emphasising their capability.
Later on, participants were shown the other two faces from the top row, and the remaining two faces from the bottom row, and they were asked to evaluate whether each person was ‘kind’, and whether each was ‘capable’.
When people were judging the kindness of new faces, they spent longer evaluating those with long hair, and when judging capability they spent longer evaluating those with short hair, regardless of the eventual decision they reached. It therefore seems that from as few as three training trials with kind long-haired faces and three trials with capable short-haired faces the participants had become sensitive to the potential link between hair-length and these traits.
Yet, when quizzed about this, no-one explicitly stated that they used such a rule. It was acquired in a non-conscious manner – influencing people’s judgements without their being aware of the source of the influence.
As a control reversed this and found the same thing in reverse
what are cues to age in a face
- facial shape
- sizes of features in the face
- texture
talk about facial shape and age from childhood to adulthood
• Pittenger and Shaw (1975) o Growth of the human skull constrained around nodal point where the brain stem meets the spinal cord. o Cardioidal (= heart-shaped) strain describes the transformation of the human cranio-facial profile as it ages. • Growth during development results in substantial age-related changes in shape information. Simply because skull shape is changing. Shape changes during adulthood are more subtle
talk about texture and age
In addition, texture (how tight and wrinkles) and colour of the face ( and hair) change with age.
who looked at surface texture and age and what did they do?
Burt and Perrett 1995)
set out to investigate the relative contribution of gross shape compared with surface texture/coloration in the perception of the age of adult faces.
• Collected together a number of different male faces spanning a range of 25 years within 7 distinct age groups
• Participants were reasonably good at judging the age of these original images.
• The different faces within an age-band were then combined into an ‘average’ face for that age-group, using the morphing techniques
• (by careful alignment of feature landmarks identified on each face, faces can be averaged together without blurring- achieved by deforming each of the images into a common average shape, in which the locations of the feature points are those of the average for the set, before blending them together.)
who looked at surface texture and age and what did they find?
Burt and Perrett 1995)
- By examining how each age-group average face differed from neighbouring averages, we can form an idea of how the faces of one age group differ from others. However, Burt and Perrett’s techniques made it possible to approach this more systematically. For instance, they were able to exaggerate the differences between age group averages to produce a computer ‘caricature’ of age-related changes
- the average face shape changes e.g. the forehead gets wider (loss of hair) chin gets broader, ears get longer.
- This ageing transform involves changes both to the face image’s shape, and in terms of its texture/ colour.
what do the presence of such a variety of cues to age help with?
The presence of such a variety of cues to age in our faces helps to make age estimation both accurate (M. Rhodes, 2009) and remarkably insensitive to changes (inversion, photographic negation, image blurring)
what are sex differences between faces
- shape information
- contrast information
Are we good at telling the gender of a face?
We are also remarkably accurate at deciding whether faces are male or female. Even if hairstyle is concealed, men are shown clean-shaven, and cosmetic cues are avoided, people are still around 95% correct at deciding whether faces are male or female.
Does classifying sex make many attentional demands?
NO- we are so good at classifying faces by sex that doing this makes few attentional demands and is little affected by having to undertake an attentionally demanding task at the same time (Reddy, Wilken, & Koch, 2004).
How can we identify what information might be used by the human visual system?
By measuring large numbers of male and female faces we can identify what information might be used by the human visual system to classify the sex of faces.
But what is important about identifying information for a face and give an e.g.
Identifying what information might be used is, however, only a first step in finding out what is actually useful or necessary for the task.
For example, one measurement that differs quite a lot between male and female faces is overall head size, since men are generally taller and broader than women. However, the size of the face is not necessarily a useful cue to determine its sex, and in experiments where faces are all standardised to the same overall size, people are no slower or less accurate at deciding their sex, suggesting that whilst head size is a cue we might conceivably use, in practice it is actually relatively unimportant for human vision.
This makes sense in everyday life, where if absolute size were to be used as a cue the visual system would have to compensate for perceived distance in order to use it
Who gathered info on physical variables that form the basis of human sex judgments and what did they do?
Burton et al.,1993) collected pictures of 91 young adult males and 88 young adult females and made a large number of different measurements of these faces. The sizes of the different features were measured in full-face images (e.g. the length of the nose or the width of the eyes), as were separations between different features. A number of different ratios were derived from these full-face measurements. In addition, profile photographs of the same people were used to recover some measurements of the 3D shape of the heads.
Give e.g.s of some of the measurements used by Burton et al., 1993
The measurements included some simple, local features such as the thickness of the eyebrows (thicker in men) and the distance between the eyes and brows (greater in women – particularly if eye-brows have been plucked), as well as more complex, 3D measures such as the protuberance of the nose
What did Burton et al., (1993) find?
On the basis of the measurements made, Burton et al. found that it was possible to classify 94% of these images correctly as male or female (a similar success rate to that achieved by human observers) using a total of just 16 different measurements
Which study gave a picture of overall differences in 3D shape for sex? What did they do?
Linney et al., 1993- can be obtained by comparing 3D surface images of male and female heads obtained using a laser-scanning technique. able to average together different surfaces obtained from a number of different male and female faces in order to produce the ‘average’ male and ‘average’ female surface. These surfaces were then compared and their differences noted.
Can be shown by using the colours of the spectrum, with red showing extreme positive differences through to violet showing extreme negative ones.
What did Linney et al., 1993 find?
Examining these images we can see that the male face has a more protuberant nose and brow and broader chin and jaw-line than the female face. The female, in contrast, has somewhat more protuberant cheeks and a fleshier pad on the chin.
Biological causes for facial differences in gender?
These differences between men’s and women’s faces have a variety of causes. For example, Enlow (1982) relates the differences in shape between men and women in the nasal area to their differing oxygen requirements. Because men are larger than women they require a greater air-fl ow, and hence a differently shaped nasal passage.
What features are used for gender (link to age)
The discussion so far suggests that, as for age, there may be different kinds of features that are useful in the perception of the sex of faces –
superficial or ‘local’ features such as the thickness of eye-brows or the texture of the skin in the beard region (where even clean-shaven men often have visible hair follicles or beard ‘shadow’),
and overall shape features such as the 3D shape of the nose and brow region.
What experiment supports the idea that not just shape information is used for sex (earlier)
Bruce et al. (1993) compared how accurately people were able to make sex judgements when shown just the 3D surface information derived from laser scanning with the accuracy obtained to the same set of faces shown in the manner in which they were scanned – with hair concealed with a bathing cap and with eyes closed.
The surface images contain all the same 3D structural information as the photographs of the people being scanned, but lack the local surface information – for example about brows and skin texture.
What did Bruce et al., 1993 find?
Accuracy with the surface images was very much better than chance (84%) but it fell considerably below that found with the photographs (94%)
Although the surface images were judged quite accurately in ¾ view, where their 3D shape could be seen, the accuracy of judging these images dropped a great deal when they were shown in full face (75% correct), while accuracy with photographs remained at 95% correct in full-face images
The local texture cues of eyebrows and stubble etc. are equally visible in full-face and ¾ photographic images, while cues to 3D shape such as nose and chin protruberance are much less visible in full-face images and must be derived entirely by an analysis of shape-from-shading
What can we learn from Bruce et al.’s 1993 study?
This shows that the additional texture information (eye-brows, visible hair, stubble and skin texture) in the photographs adds significant information to that contained within the 3D surface alone
what else but texture info and shape is important
contrast information
talk about contrast information and gender in faces
Russell 2009- we don’t only use shape information there is an additional judgment.
• Creation of shape-androgynous faces via morphing- created an average of a number of both male and female faces then morphed from one end pint to other and picked face in middle (androgynous face).
• Changed facial contrast- greater contrast in the more female looking image
• Even if you have the same facial shape the face can be more male or female.
• Faces with androgynous shape can be made to appear more female or male, simply by manipulating contrast
• Pigmentation information is important.- surface information contributes to perceived sex
can make an androgynous face be made to look more like a man or more like e a woman simply by decreasing (more masculine looking) or increasing (more feminine looking) contrast in these regions.
