ÐÓ°ÉÔ­´´

Perception, learning and remembering, evaluating information, selecting
and making responses – our brains continuously carry out these functions
and more, apparently without much effort. But for some people these activities
suddenly become difficult and laborious as a result of a stroke or head
injury. Studies of individuals who have suffered brain damage can give valuable
insights into how the healthy brain functions.

Neurologists have known since the 19th century that brain disease does
not necessarily result in an across-the-board reduction of abilities. Certain
functions might be severely affected while other skills remain unimpaired.
A German neurologist, Hermann Wilbrand, observed in 1892 that one of his
patients could see and read but had severe problems recognising objects
and people. She confused her doctor for a dog, and the maid for a table.
During the same year a French neurologist, Jean Dejerine, described a patient
who could recognise people or objects, but could no longer read.

These observations suggest that the two abilities – reading and recognising
objects – are at some point carried out by different structures in the brain.
This is not to say that these abilities are separate from beginning to end.
Both reading and recognition of objects rely on visual input and both could
produce the same response – for example, saying the name of the object or
reading aloud. To use the anology of information processing, the idea is
that some of the processes underlying these abilities are arranged in a
parallel fashion.

Many neuropsychologists use a ‘modular’ concept of brain functions.
The idea is that information is processed through an array of distinct but
interconnected mechanisms, each of which performs a certain operation. The
advantage of modular organisation is that damage to one part does not necessarily
affect the function of the rest of the system. It is like a hi-fi system
consisting of a number of interconnected modules: radio, amplifier, loudspeakers
and so on. If, say, the cassette recorder is broken, one can still play
records.

The existence of independent faults in processing different aspects
of visual information clearly fits in with the modular notion. People can
experience highly selective impairments, such as an inability to perceive
different tones of grey, movement or spatial position. This does not explain,
though, the full nature of visual experience. When we look around we see
an orderly arrangement of objects in three-dimensional space, and we are
not aware that, say, colour might be determined independently from form.

Defining awareness or consciousness is a long-standing philosophical
problem. Neurologists have traditionally made a distinction between the
conscious state (for example, a person in a coma is not conscious) and the
contents of conscious experience. The American philosopher Daniel Dennett
stresses the notion of personal access to these contents. Your brain might
be processing a lot of different things at the same time (the word you are
reading, the temperature in the room and so on), but you are not consciously
aware of all this information simultaneously. Instead, there is a conscious
access to only some of the available information. Different ‘impairments
of awareness’ in people with brain damage provide glimpses into how the
healthy brain directs and maintains awareness.

Some people find it difficult to recognise faces, although they can
read and recognise objects. The condition is known as prosopagnosia from
the Greek prosopon (face) and agnosis (without knowledge). Prosopagnosic
people know when they are looking at a face, can identify and describe separate
features such as the eyes and mouth, and can even determine age, gender
and expression. But they cannot tell when looking at a person or at a photograph
if the face is one they have seen before. In severe cases, individuals will
fail to recognise family and close friends, or their own face in a mirror.
Not even a vague feeling of familiarity survives.

In 1984, Russell Bauer at the University of Florida published a surprising
study of a man who was completely prosopagnosic. In tests, he could not
put the right name to a photograph of a familiar face when given a choice
of five alternative names. Bauer, however, simultaneously recorded the level
of electrical conductance of the man’s skin. This is a sensitive measure
of the amount of sweat produced by the skin, which is in turn an indication
of activity in the autonomic nervous system – the section of the nervous
system that operates involuntarily and controls vital bodily functions such
as heart rate and sweating. Changes in skin conductivity happened more often
and with higher amplitude when the examiner spoke the correct name than
for the four foils. Antonio Damasio and his colleagues at the University
of Iowa also found increased responses in skin conductivity when their prosopagnosic
patients looked at slides of familiar faces embedded among those of unknown
persons.

Even though these people could not recognise familiar faces, they must
have done so unconsciously in some way, given that their autonomic nervous
system responded differently to familiar and unknown faces. This phenomenon
of recognition without awareness is called ‘covert’ recognition. ‘Overt’
recognition, in contrast, refers to recognition abilities to which one has
conscious access.

At the Neuropsychology Unit of The Radcliffe Infirmary, Oxford, our
idea was that face recognition in the healthy brain is carried out by a
sequence of subprocesses or modules, some of which are not open to introspection.
Although some of these processes might continue to function in people with
certain kinds of brain damage, they are unaware of this because mechanisms
later in the sequence are damaged, disconnecting the intact processes from
awareness.

We set out to test this notion, working with Andrew Young of the Department
of Psychology at the University of Durham. Our approach was to use tests
which involve familiar and unfamiliar faces, but which do not require an
overt identification of the faces used. The first series of experiments
was with a patient called PH, who was unable to recognise faces after suffering
head injuries in a road accident.

First, we wanted to rule out the possibility that he had some vestigial
awareness of familiar faces but lacked the confidence to acknowledge familiarity.
This we checked by a method known as forced-choice guessing. Over a large
number of trials, we presented PH with two faces: one familiar and the other
unfamiliar. We asked him to point to the one that appeared familiar, and
to guess if he was unsure. He was correct on 51 per cent of the trials –
a chance performance. It seemed safe to conclude that PH was unable to overtly
extract any information regarding identity from photographs of familiar
faces.

The first experiment aimed at detecting covert face recognition involved
‘matching’ – deciding whether two photographs of faces, presented together,
are of the same person or different people. This task does not call for
conscious recognition of the face per se. And if both faces were of the
same person, we used photos taken from different angles to avoid the possibility
of pattern-matching rather than person-matching strategies. PH responded
as normal subjects do: he could match photographs of famous or familiar
people faster than those of unfamiliar people, even though he could not
identify any of the faces.

The second set of tests made use of an ‘interference effect’. We showed
PH a series of faces, each accompanied by a name printed in a speech bubble.
The name could be that of the person in the photograph, of a person in a
related occupation or of a person in an entirely unrelated occupation. The
task was to decide as quickly as possible whether the printed name was that
of a politician or not.

Normal subjects are influenced by the accompanying face: if the printed
name is unrelated to the face, they take longer to respond correctly than
for the other two conditions. PH, who had no difficulty judging whether
a printed name by itself was that of a politician, showed the same interference
effect: he responded faster to the printed name when it was attached to
the correct or unrelated face than when it was accompanied by an unrelated
one.

Covert processing can also be demonstrated in ‘associative priming’
experiments. Such experiments are designed to show the effect that a ‘prime’
stimulus has on the response to a subsequent ‘target’ stimulus. In this
case, we showed PH a picture of a face (the priming stimulus), followed
quickly by a written name (the target stimulus). The response required was
to decide whether the name was familiar. If there is a strong association
between the prime stimulus and the name, normal subjects will respond faster
than when the prime is an unfamiliar person, or a familiar person not normally
associated with the name.

PH also showed this type of priming effect from faces that he did not
overtly recognise. Moreover, we could compare the priming effect from face-primes
with that triggered by name-primes (which PH recognised without difficulty).
The response times were the same: overt recognition of name-primes did not
reduce the time PH took to respond to the name-targets. This suggests that
PH not only covertly recognised faces, but that his recognition was ‘normal’
in the amount of associative priming that it produced.

PH’s preserved abilities on these tests appear very similar to these
aspects of recognition that seem to operate automatically in normal people.
We believe that his covert face recognition stems from an intact fact recognition
system which has become disconnected from the processes that lead to an
awareness of recognition. Some of his face-recognition modules are preserved,
but he has no conscious access to the outputs.

We have subsequently tested other patients suffering from prosopagnosia
and found signs of evidence for convert recognition in some of them, but
not all. In one individual, who has difficulty recognising objects as well
as faces, we failed to find any signs of convert recognition will lead to
a much more sophisticated taxonomy of the effects of brain injury.

Other researchers have made similar observations of ‘knowledge without
awareness.’ For example, Tim Shallice at the Department of Psychology, University
College London, has studied people with reading difficulties. They can no
longer read words ‘at a glance’ but have to resort instead to a slow and
painful strategy of identifying the individual letters. Nevertheless, they
can respond to words presented for such brief exposures that even unimpaired
people would not be able to read them. Shallice found one patient could
tell whether such a stimulus was a word or a nonsense string of letters
and, even more surprisingly, whether a printed name was that of an author
or a politician. Subliminal effects, where the stimuli are presented for
such a short time that recognition is impossible, are also found in healthy
subjects. But the effects are all the more intriguing in brain-damaged subjects
who, in other circumstances, cannot recognise such a stimulus.

Remember without remembering

Studies of memory disorders (amnesia) have also provided examples of
‘covert knowledge’, in that it seems that amnesiac patients can sometimes
remember without being aware that they are remembering. These include some
classical neurological anecdotes. For example, at the turn of the century,
the eminent Swiss psychologist, Eduard Claparede, secreted a pin in the
palm of his hand just before shaking hands with a patient suffering from
severe amnesia. At their next encounter, she was reluctant to shake hands
with Claparede but was at a loss to say why.

Of the many accounts of amnesiac patients who can acquire new skills
the most famous is HM. Brenda Milner and her colleagues at the Montreal
Neurological Institute have studied HM extensively for the past three decades.
His amnesia was so severe that he could not remember any words or pictures
after even a short delay. An experimenter who left the room for only a few
minutes would, on return, appear completely unfamiliar to him. Yet HM was
able to learn motor skills, such as tracking a moving light with a pen,
even though he had no conscious recall of the training sessions.

Other researchers have found evidence for preserved covert learning
using other indirect tasks. Elizabeth Warrington of the Natinal Hospital,
London, and Lawrence Weiskrantz of the University of Oxford used a series
of pictures of an object. The object was initially fragmented, and then
displayed with the contours increasingly filled in until the object was
complete. After repeated testing, their amnesiac patients could recognise
the pictures at a more fragmented stage although they had no recollectin
of having previously seen them. Similarly, Nelson Butters and his colleagues
at the Veterans’ Administration Medical Center, San Diego, have demonstrated
implicit learning of words in amnesiac patients.

Another aspect of awareness is revealed in neurological patients who
are unaware of their impairments (anosagnosia) and fail to comprehend their
problems, or even deny them. Anosagnosia is not necessarily a global lack
of awareness; it may be restricted to one domain or to one modality of behaviour.
In a seminal study, published by the neurologist G Anton in 1899, a man
with blindness caused by damage to the cerebral cortex would deny any suggestion
that he had lost his sight, and would attempt to navigate his wheelchair
through the ward. Yet this same man was quite prepared to admit his other
symptoms, including paralysis of one side of his body.

We have recently worked with a woman who, after a severe stroke, was
left with constellation of cognitive and motor impairments: partial paralysis
on the left side of her body, loss of one half of the field of vision in
both eyes, severe memory problems nad a marked inability to recognise faces.
She was painfully aware of all her problems except for her difficulty with
faces. Each time she failed to recognise a photograph of a famous media
celebrity (some of whom she knew personally), her problem was pointed out
to her, and each time she would react the same bland, disbelieving manner.

So not only loss of awareness but also denial of loss can be confined
to a particular aspect of cognitive or motor function. We have to envisage
brain mechanisms that work to maintain awareness and that monitor processes
which are automatic and hidden from conscious inspection. From the evidence
that impairments of awareness can be highly selective, one can deduce that
this regulation of awareness is to some degree decentralised in the brain
rather than confined to one area.

Daniel Schacter of the University of Arizona was incorporated the theoretical
implications of these findings in his current model of brain functioning.
In essence, he envisages a conscious awareness system that receives inputs
from a variety of knowledge sources, or modules. We can then ask whether
the module itself (say, the face-recognition system) is destroyed, degraded
of disconnected from the conscious awareness system. If we can devise more
tests to identify covert and overt processing, we should be able to address
this question in individual patients.

In practical terms, it may be possible to harness covert recognition
and covert knowledge to retrain disabled patients. Schacter and his colleagues
have shown that rehabilitation based on systematic and repetitive instruction
can be effective in patients who are not consciously aware of previous training
sessions. For instance, they taught severely amnesiac people to operate
a computer and to understand some computer language. They adopted the painstaking
method of vanishing cues. In the course of learning which word (for example,
‘loop’) to attach to a particular operation, the letters are successively
eliminated until the trainee can remember the correct command from the cue
‘l’ Some amnesiacs even learn to operate the computer without the need for
any cues.

It is intriguing to speculate why vanishing, partial or fragmented cues
appear to have this privileged impact on recognition and learning. Is it
the fact that the person is required to make a particular effort to decipher
the partial cue? Or do the partial cues turn the task into one which is
somehow more appropriate to the patient’s intact skills? We are far from
understanding the mechanisms of preserved covert learning and recognition
skills. It is rewarding and stimulating, however, when science and clinical
practice interleave to the benefit of both theory and therapy.

Edward De Haan is a research fellow and Freda Newcombe is director of
the Russel-Cairns Head injury Unit at The Radcliffe Infirmary, Oxford.