Chapter Five
Model 3:
Conscious experience is Œinformative‹ ---
it always demands some degree of adaptation.
"Into the awareness of the thunder itself the awareness of the
previous silence creeps and continues; for what we hear when the
thunder crashes is not thunder Œpure‹, but thunder-breaking-upon-
silence-and-contrasting-with-it."
--- William James, 1890 (p. 156)
5.0 Introduction: information and adaptation.
5.1 The adaptation cycle: Any learnable task goes from Œcontext-
creation‹ to conscious Œinformation‹ to Œredundancy‹.
5.11 Context-creation.
5.12 Conscious experience corresponds to the information
stage of adaptation.
5.13 Redundancy Effects occur at all levels of processing.
5.14 Adapted systemscan provide new context for later
conscious experiences.
5.2 Human beings also Œseek‹ information at all levels.
5.21 Perceptual systems aim for informative parts of the
field.
5.22 Conceptual processes aim for informative points.
5.23 Goals define significant information to attend to.
5.24 Summary: Seeking Œversus‹ adapting to information.
5.3 Model 3: Interpreting informativeness in the theory.
5.31 How does the system know what is informative?
5.32 It takes time for GW input to become conscious.
5.33 When one conscious event becomes redundant, the next
most informative input becomes conscious.
5.34 Objectification and contextualization.
<j <†5.4 When repeated experiences do not fade: Is informativeness a
Œnecessary‹ condition for conscious experience?
5.41 The apparent implausibility of Redundancy Effects
in everyday life.
5.42 Conscious fading can be countered by a continuing
mismatch between input and context.
5.43 Summary.
5.5 Implications for learning.
5.51 Conscious experiences trigger widespread adaptation and
learning.
5.52 Learning alters the experience of the material learned.
5.53 Is consciousness necessary for learning?
5.6 Some experimental predictions.
5.7 Other implications.
5.8 Summary.
<, <
5.0 Introduction: information and adaptation.
Has the "publicity metaphor" helped to clarify the issues so
far? Consider how the GW model resembles a publicity organ like a
newspaper. First, a global workspace can distribute a message to
a wide public of specialized, relatively independent processors
(2.x). Further, only one consistent message can be broadcast at a
time, so that the mental news medium does not publish self-
contradictory information (2.x). And third, GW theory claims that
there are contextual constraints on the conscious interpretation
of the world, comparable to the editorial policies and practices
of a newspaper that determine how it will select and interpret
the news (4.0). In these ways the publicity metaphor seems
helpful.
But we have not yet addressed some essential features of the
publicity metaphor. ŒFirst‹: So far, we cannot tell old from new
input; ŒSecond:‹the model has no way to determine the Œsignificance‹
of a piece of news; And Œthird‹: we have no way to keep the system
From publishing the same old message over and over again.
short, until now we have a newspaper that has no preference for
news.
Yet the nervous system does have a great preference for
news. There is ample evidence from a great variety of sources
that people and animals actively seek novelty and informative
stimulation, and that they have an enormous selective preference
for significant input (5.13). Repetitive stimuli tend to fade
From consciousness regardless of their sensory modality, degr
of abstractbess, or physical intensity (short of the pain
threshold) (5.1, 5.2). Even a single neuron habituates to
repetitive input, and becomes active again only when the input is
changed (Kaidel, Kaidel & Weigand, 19xx). The GW system is
designed especially to cope with novelty and informative
stimulation, because it allows many knowledge sources to work
together on a single, novel source of information. The primary
function of consciousness, we will argue, is to facilitate this
cooperative integration of novel information (11.0). The more
informative an event is, the more adaptation is required, and the
longer the event must be in consciousness for adaptation to be
achieved (5.5).
ŒDefining "information."‹
In this chapter we will use the word Œinformation‹ in its
conventional sense as a Œreduction of uncertainty‹ in a set of
choices defined within a stable context (Shannon & Weaver, 1949;
Miller, 1953). The context of information must define at least<j <
two options: 0 or 1 in the case of a computer, or "war" and
"peace" in the case of a diplomatic code. Any arbitrary amount of
information can be coded as a combination of binary codes. This
is of course the well-established mathematical definition that
has been so influential in communication engineering and computer
science, except that we will be using it qualitatively, and in a
somewhat broader sense. Over the past few decades, the
mathematical definition has also found increasing application in
psychology. It has been found useful in modeling fundamental
findings about reaction time (Hick, 1952; Hyman, 1953), classical
conditioning (Rescorla & Wagner, 1972), basic level categories
(Gluck & Corter, 1985), perceptual complexity (Garner, 1974),
etc. Thus the mathematical notion of information seems to have
some psychological reality.
How is information in this sense related to consciousness?
There is good evidence (presented below) that we are conscious of
an event only when it exists in a stable context, but not when it
is so predictable that there are no conceivable alternatives to
it. Conscious experience seems to exist only when there are some
degrees of freedom within a stable context. Thus the notion of
reduction of uncertainty in a stable context seems appropriate.
Information is inherently context-dependent, and we have
presented a set of arguments before (2.xx, 4.xx) that
consciousness is also highly context-dependent.
Conscious experience of the world is not a direct function
of physical stimulation. The same physical stimulus repeated over
and over again will soon become less informative --- and also
less conscious. But a highly significant or variable stimulus
habituates more slowly. We therefore need to make a distinction
between Œphysical stimulation‹ and real Œinformation‹. On the other
side of that coin, this is the difference between Œrepetition‹ and
Œredundancy‹. The same physical energy impinging on the same
sensory receptors may be either informative or not, depending
upon the reduction of uncertainty in the relevant context.
Sometimes the physical Œabsence‹ of an expected stimulus can
provide information, just as its presence may be redundant. In
this sentence, we need only omit one ... to show that the absence
of a stimulus can draw our attention --- and the missing item may
well become conscious for the reader. Thus information and
stimulation are not the same; they can vary independently. In
general, the probability of being conscious of any event
increases with its information value and decreases with
redundancy.
Finally, the same stimulus can carry different amounts of
information when it suggests something beyond itself. In Pavlov's
conditioning experiments, when the sound of the bell signaled
that food was coming (a significance beyond itself), the hungry
dog was much more likely to prick up its ears, the orienting
response to the bell took longer to habituate, and learning
occurred more quickly. One way to think about significance is in
terms of purposes the hungry dog is likely to have, which create<j <
Œgoal contexts‹ for its perceptual systems to explore (4.23).
Significant information can then be seen as a reduction of
uncertainty within a goal context (5.xx). Thus the concept of
information can be related naturally to the things that matter
most to an animal or human. We can think of information as
existing at different levels, just as we have previously
suggested that contexts exist at different levels (4.2).
The strongest argument for the close relationship between
information and consciousness is the existence of what we will
call Redundancy Effects. Redundancy, the absence of information,
is defined in information theory as the transmission of a signal
after the uncertainty at the receiver is already zero. The choice
between "peace" and "war" had great information value in 1945 for
most of the world, but repeating the word "peace" over and over
again after that point became increasingly less informative, even
though the context of subsequent events is accurately described
by that word. Thus the word "peace" became increasingly
redundant, but not false. There are many well-known cases in
which conscious input fades with repetition --- cases like
stimulus habituation, automatization of skills and mental images,
perceptual adaptation, shifts in the Adaptation Level of
perceptual and judgment categories, "blank-outs" in the Ganzfeld,
semantic satiation, loss of access to stable conceptual
knowledge, etc., etc. These phenomena allow us to do a
contrastive analysis, showing a direct relationship between
conscious experience and the informativeness of an event (Table
5.13).
Habituation of awareness to a repeated stimulus is the most
obvious example of a Redundancy Effect (Table 5.13). At this
moment the reader is likely to be habituated to the feeling of
the chair, the color and intensity of the ambient light and
background sounds, the language of this book, and many other
predictable features of the inner and outer world. A previous
chapter (1.xx) detailed Sokolov's arguments for the continued
existence of Œun‹conscious representations of habituated stimuli.
Sokolov argued a mismatch in any parameter of a habituated
stimulus will elicit a new Orienting Response. To detect such
mismatches, we must maintain some sort of representation of the
expected input. But this representation does not elicit an
Orienting Response --- or, in the terms used in this book, it is
not normally conscious. Thus there must be an unconscious
representation of a habituated stimulus which is similar in many
respects to the conscious perceptual representation of the same
stimulus when it first occurs. The Sokolov argument therefore
allows us to contrast two stimulus representations under
identical physical conditions: the conscious representation which
occurs when we first take in a stimulus, and the representation
that continues to exist after habituation (Table 5.13).
<j <† ŒInformation vs. novelty (mismatch)‹
What can be the difference between the conscious and
unconscious representations of the same stimulus? Several writers
have suggested that Œnovelty or mismatch with expectations‹ is
involved in consciousness perception. It is sometimes maintained
that there must be a mismatch between input and expectations for
a stimulus to be conscious. This is certainly true in the case of
surprise, as discussed above. But it cannot be true without
qualification (Berlyne, 1960). Any stimulus that violates
previous expectation can only do so in a context which is itself
not violated --- if input were to violate every expectation, if
it were Œtotally‹ new, it could not be experienced at all.
Therefore all understandable novelty exists within a relatively
stable context that is not novel.
The opposite argument has also been offered. Marcel (1983
ab) suggests that a match between input and memory is required
for perceptual input to be conscious. But this cannot be
completely true either: if there is a perfect match between input
and expectation, we have the case of habituation and loss of
consciousness. We find ourselves in a middling position with
respect to both the match and the mismatch hypothesis: yes, there
should be some sort of match, but not too much. Both the mismatch
and the match hypothesis capture some, but not all, of the
evidence.
We will develop the argument that the notion of information
is more attractive than either simple match or mismatch.
Information involves both a match of context and a mismatch of
the stimulus details. It further suggests that the input must be
useful, in the sense that many systems can reduce their
uncertainty relative to it. It also allows us to include the
notion of significance as a reduction of uncertainty in a
relevant goal context. And it seems to explain the
well-established Redundancy Effects.
If the concept of information is indeed preferable, what
about the case of surprise, which is indeed a mismatch of input
and expectations? Mismatch reduces to a special case of
information --- it is the case where the context of the expected
input is itself violated. This context then becomes
"decontextualized" (4.xx), and its components must be reassembled
in the stable framework of a higher-level context. We have
discussed this case in some detail in Chapter 4. The point here
is that the notion of information seems wellcsuited to handle a
number of important properties of conscious input; it can also
explain mismatch phenomena like surprise.
< ŒSome possible counter-examples.‹
Much of the argument depends upon the Redundancy Effects,
those cases where repetition leads to a loss of conscious
experience. There are some apparent counter-examples to the
Redundancy Effects: cases where repeated events do not fade, or
where they seem to become Œmore‹ consciously available with
practice (5.4). For instance, conscious access to highly
significant or unpredictable events is lost only slowly. In the
case of chronic pain people do not lose conscious access at all.
We suggest that these facts reflect the special role of
signficant information. But as we have mentioned, significance
can be treated as a reduction of uncertainty in a higher-level
goal context (5.xx).
Further, there are cases in which practice seems to Œincrease‹
access to conscious events. For instance, practicing recall will
bring memorized material to mind more readily (Erdelyi, 1985),
and practicing visual search will cause the search target to
"pop" into consciousness quite involuntarily (Neisser, 1967;
Shiffrin & Schneider, 1977). Notice however, that what is being
practiced here is not the visual or memory target, but the
Œprocess of search or recall‹. That is, in these cases we gain
automaticity in the skill of controlling Œaccess‹ to consciousness
(viz., Chapter 8), but the input that will become conscious is
not predictable, and may be quite novel and informative. Thus,
these facts do not contradict the claim that consciousness
requires informative input. Indeed, the process of recall or
search itself does become automatic and unconscious with
practice. Only its result remain informative and conscious.
There are also cases where repeated stimuli fade from
consciousness, only to return in a new guise (Warren, 1961, 1968;
Pritchard, Heron, & Hebb, 1960). As we shall see in section 5.4,
these apparent counter-examples can be readily understood with a
deeper understanding of information and redundancy. Namely, these
cases seem to reflect the fact that the same input can still be
informative if the context of interpretation changes. There is
indeed evidence for this suggestion from a variety of sources.
From the viewpoint of information theory, a change in conte
does create new information, even with repetitive input.
We conclude that Redundancy Effects are both powerful and
pervasive, while apparent counter-examples can be explained
plausibly in an extended information-theoretic framework. All
this supports the idea that informativeness may be a Œnecessary
condition‹ for conscious experience of some event. This viewpoint
also suggest a new perspective on context (4.0): In a sense,
context consists of those things to which the nervous system has
Œalready‹ adapted; it is the ground against which new information
is defined.
<j <†
ŒA terminological note.‹
Our use of the term "information" is similar to the
classical mathematical definition developed by Shannon and others
(Shannon & Weaver, 1949), but we should make note of some
possible differences. Psychological contexts are highly complex
knowledge structures that have many more dimensions than the
simple, binary, one-dimensional message contexts of classical
information theory. But of course, we know that knowledge
structures of any dimensionality and complexity can be reduced in
principle to binary choices (Shannon & Weaver, 1949). Further,
the classical definition presumes that reduction of uncertainty
takes place in a stable context of choices; but psychologically,
we know that contexts are not totally stable, but adapt to
informative input whenever possible. The nervous system learns
about predictable inputs; it is not passive like the contexts of
conventional information theory. We will argue below, however,
that conscious experience is associated with a range of phenomena
in which the context of informative choices is Œrelatively‹ stable.
Within these relatively stable contexts, the classical definition
is quite useful. Finally, the formal definition of information
is quantitative, but we will not develop a quantitative approach
here. Quantification at this stage can apply only to a small,
experimentally defined subset of the full range of phenomena.
Others have already done this (see references cited above). We
will focus here on the making a case for the special relationship
between consciousness and information in general.
ŒAdaptation‹
After information, the second major concept in this chapter
is Œadaptation‹. Here we will use it in a narrow sense, as the
process of learning to represent some input --- to know and
understand it to the point of automatic predictability. Learning
to represent something involves, of course, a reduction of
uncertainty (i.e., information). When there is a perfect match
between input and its representation, the input is Œredundant‹ with
respect to its representation. Thus redundancy is the end-product
of successful adaptation.
We can borrow Piagetian terms here to represent different
ends of the adaptation continuum (Piaget, 1952). When confronted
with a situation that is new and strange, people need to find new
contexts for experiencing the input; the result resembles
Piagetian accomodation. In other words, accomodation has to do
with the discovery of usable contexts. On the other end of the
continuum, when the input is highly familiar and predictable,
minimal adaptation is required, so that we can assimilate it into
readily available contexts. In the extreme case of redundancy,<j <
context and input match exactly.
Conscious experience of an event seems to occur midway
between the outer poles of assimilation and accomodation. If we
can automatically predict something completely, we are not
conscious of it. But if the input requires a deep revision of our
current contexts, we do not experience it either --- it is too
confusing or disorganized to experience Œas such‹, though we may
experience fragments and tentative interpretations of the input.
Somewhere between these two extremes, between the assimilation
and accomodation poles of the continuum, we may have an accurate
conscious experience of the event. From the adaptation point of
view, an informative conscious event translates into a demand for
adaptation. This is of course the claim stated in the chapter
title: that conscious experience is informative --- it always
demands some degree of adaptation.
In sum, there is a close web of connections between certain
fundamental ideas: information, consciousness, reduction of
uncertainty, a drop in contextually defined alternatives, a
demand for adaptation and learning, a gain in predictability and
redundancy, and the creation of new contexts.
Adaptation takes place over time, and we develop now the
notion that conscious experience corresponds to a certain stage
of the "adaptation cycle" --- namely, the stage where there is a
relatively stable context for understanding the input, but there
is still uncertainty to be reduced within that context. Many
processors can cooperate in reducing the uncertainty. A
fundamental point is that the nervous system is always in dynamic
adaptive activity --- it is always working to adapt to conscious
input --- even when we Œseem‹ to be conscious of only a single
thing. Conscious experience cannot be understood apart from this
process of dynamic adaptation. We turn to this issue next.
5.1 The adaptation cycle: Any learnable task goes from Œcontext-
creation‹ to Œconscious information‹ to Œredundancy‹.
In learning about a new source of knowledge we often start
with considerable uncertainty and confusion. By paying attention
to the problem, a sense of clarity is often gained, as we become
more conscious of what is to be learned. Finally, with practice,
the material becomes highly predictable and fades from
consciousness. These three stages make up what we will call the
Œadaptation cycle‹: starting only with the knowledge that there is
something to be learned, the first stage of Œcontext creation‹ is
resolved as the elements to be learned are defined; in the second
stage we have a working context for understanding the new
material, which is now Œinformative‹ --- that is, input now serves<j <
to reduce uncertainty within the working context. In the third
stage we have Œadapted‹ completely, and lose conscious access to
the learned material. Consciousness is primarily involved in the
first two stages, but in the first, the input is so ill-defined
that we are not even truly conscious of it Œas such‹. Consciousness
of the input as such is confined to the second stage, which we
call the stage of informativeness.
Below we present a number of empirical findings that
support these points.
5.11 Context-creation.
-----------------------------------
Insert Fig. 5.11 about here.
-----------------------------------
Consider Figure 5.11, which looks at first to most people
like a random collection of black and white spots. It is in fact
a coherent picture of something; but in order to experience it we
need some context. Some may be provided by the picture's title,
"Dalmatian in the Park"; some percentage of obervers will find
this hint helpful. (Note that if this helps, you may not be
conscious of the title Œas such‹ at the moment it seems to help ---
i.e., the effect is contextual, since influences that are
unconscious at the moment in question help shape the conscious
experience.) Other observers find that it helps to notice that
the diagonal black "lines" converging toward the center are the
edges of a sidewalk in a park. Knowing this may help to
reconstruct the three-dimensional properties of the picture. But
once again, depth information created by the converging sidewalks
will be unconscious at the moment when it constrains and helps
reveal the conscious object. Other conscious hints that help to
create context include the dog's nose, the tree above, the
circular planter in which the tree stands, and the realization
that a black-and-white photograph of a spotted Dalmatian in a
shadow-flecked park can indeed look like this.
A good deal of learning begins in confusion. Until the
confusion is dispelled, until guidance is provided either by the
material itself, by a parent, guide or teacher, or by the use of
knowledge and strategies from the past, we do not fully
experience the material that is to be learned. This point is not
limited to perception. At the end of section 4.13 we used a
confusing paragraph about washing clothes, but without mentioning
the topic. The is point that linguistic topics are often
ambiguous. Providing a title ("washing clothes") creates enough
context for the paragraph, so that we can become conscious of its
meaning as a whole, rather than as a fragmented and incoherent
set of words and sentences.<j <†
Often we have only a goal context in confronting new
material. Someone tells us that here is something interesting or
important --- pay attention to it and you will become aware of
it. This is how the reader was guided in the Dalmatian
demonstration above. It is how psychologists usually get people
to do things in experiments. We tell subjects which goals to
pursue, and see how they do: "Please pay attention to this, try
to memorize that, tell us what you see there." Even just
providing a goal context helps narrow down the possibilities.
Of course, the context of the experience itself is evoked by
conscious events (4.xx). Context-creation may involve tentative
conscious interpretations of the input, or conscious fragments,
or consciousness of a different level of analysis than the
desired experience. In the Dalmatian demonstration, the reader
was surely conscious of black and white spots even at the
beginning. But in the stage of context-creation one is not
conscious of the material to be learned Œas such‹. The Dalmatian
becomes conscious only after context has been created for it.
5.12 Conscious experience corresponds to the information
stage of adaptation.
Harking back to the Dalmatian as a case in point (Figure
5.1), once we have the appropriate contextual constraints for a
figure, we can evidently become conscious of it. But what does
that mean?
Inherent in the notion of conscious experience is the
existence of features of the conscious event. The Dalmatian has
size, location in space, color, texture, and so on. But features
involve discriminations of some sort: at the very least judgments
of presence or absence, and implicit comparisons to other
features. The Psychophysical Law implies that perception of
intensity always involves an implicit comparison to previous
intensities (Stevens, 1966), and this point is not limited only
to intensity. Many aspects of a perceptual event, such as color,
brightness, and pattern, are known to involve implicit
comparisons (e.g., Rock, 1983; Uznadze, 19xx; Garner, 1974).
Judgments of conceptual qualities are also thought to involve
implicit comparisons (Helson, 1964). Research on person
perception clearly shows that we perceive people in comparison to
others (Weiner, 19xx). In sum, conscious experiences can often be
shown to involve not one, but at least two alternative
representations, one of which is implicit. Whenever we become
conscious of something, internal processes are saying, in effect,
"Aha! It's a dog with black and white spots, not a brown dog, or<j <
a cat, or some other object." Needless to say, this kind of
implicit comparison must take place at many levels of analysis at
the same time.
Another way of saying this is that conscious events are
Œobject-like‹: they have many correlated features, which are
implicitly differentiated from alternatives. This is quite
different from habituated representations, which are not
experienced as object-like. Since habituated representations are
highly predictable, we may presume that their alternatives are
represented, if at all, as highly unlikely. In a later chapter,
the object-like nature of conscious experience will become very
important (9.0).
A direct connection between conscious events and
quantitative measures of information has been established in
several cases. Take for example the notion of "basic level
categories." In a number of studies Rosch and her colleagues have
shown that people tend to think of the world in terms of
prototypical members of categories. Thus we think of "furniture"
not so much as an abstract class, but in terms of its typical
members like chairs, tables, and beds; other members like
ashtrays, drapes, and vases are much less likely to be thought of
spontaneously (Rosch, 1975). Further, objects fit into a
hierarchy of abstraction, with the most typical objects occupying
a middle level in the hierarchy between great specificity and
great generality. Thus a ŒColonial kitchen chair‹ is quite
specific; Œkitchen chair‹ is more general, followed by Œchair,
furniture, artifact, inanimate object‹ and Œphysical object‹. The
word Œchair‹ has been thought to be the most typical member of the
hierarchy: it is easier to describe, recognize, draw, recall, and
the like (Rosch, Mervis, Gray, Johnson, & Boyes-Bream, 1976).
Typical objects are probably easiest to bring to consciousness.
However, there is now evidence that the level of an object
hierarchy that is easiest to use at any given time depends upon
the alternatives that are being entertained. If we are
considering the esthetic pros and cons of man-made objects vs.
natural scenery, "chairs" are not necessarily the best example of
man-made objects. Similarly, if we are debating the merits of
Colonial vs. modern kitchen chairs, very specific differentiating
features are likely to come to mind. Depending on our purposes
and implicit comparisons, different levels of the hierarchy of
abstraction are likely to come to mind. Along these lines,
Barsalou (1983) and Murphy & Medin (1985) have shown that
conceptual structures are highly unstable and vary with the
context of alternatives. Gluck and Corter (1985) have developed a
set of quantitative predictions based on this reasoning. They
have modified the well-known mathematical formula for information
to measure Œrelative informativeness‹, and find that the resulting
model accurately predicts which level of a category hierarchy
will be chosen under different circumstances. Gluck and Corter
suggest that "the most useful categories are those that are, on
the average, optimal for communicating information (hence<j <
reducing uncertainty) about the properties of instances."
There are other connections between likely conscious
contents, implicit comparisons, and mathematical information
theory. Garner (1974) has shown that people prefer to select
stimuli that are neither very high nor very low in information
content, suggesting that we tend to pay attention to events that
do not require enormous adaptation, but that do require some.
There is also good evidence that choice reaction time, the time
needed to choose between explicit alternatives, is a function of
the mathematical bit-rate, the quantitative measure of
information (Hick, 1952; Hyman, 1953). Even classical
conditioning is a function of the amount of information carried
by the signalling stimulus (Rescorla & Wagner, 1972).
In communication, we select the most informative features to
convey to the listener ccc that is, the features with the most
valuable comparison. We have already mentioned Olson's revealing
experiment in children's speech, in which the context of
alternatives determines whether the child will say, "The gold
star is under the Œsquare‹ block" or "The gold star is under the
Œwhite‹ block" --- even though the block in question is the same
one (Olson, 1970). But we can even find this reasoning going back
to Aristotle, who defined a "definition" as a statement of a
general category plus a differentiating feature in the category.
"A mammal is an animal that suckles its young." "A chair is a
piece of furniture for people to sit on." The general category
provides a context of alternatives, and the differentiating
feature reduces those alternatives to the correct subset.
This pattern is fundamental in linguistics. Conversations
depend upon the "given-new" contract, the shared understanding
that certain things are true, so that they do not need to be
repeated, allowing new information to be brought to the fore
(Haviland & Clark, 1974; Clark & Carlson, 1981). Similarly,
individual sentences have a "focus" and a "presupposition": In
the sentence, "It was the book that I lost yesterday", "the book"
is brought forward in the sentence and made more focal. For
comparison, "Yesterday I lost the book" does not have this
special focus. There are numerous linguistic techniques for
bringing some information to the fore, and backgrounding other
messages that are already shared. When hearing a sentence, people
seem to pay attention primarily to whatever is new, focal,
topical, and emphasized (Chafe, 1974). There is implicit deTemphasis of anything that
is known, peripheral, and irrelevant in
the moment.
If conscious events indeed exist in a context of implicit
alternatives, how do we know that these alternatives are being
reduced? After all, information is defined as a Œreduction‹ of
alternatives. Part of the answer depends, of course, on the
Redundancy Effects discussed below. But consider one of the most
obvious aspects of conscious events: the fact that they are so
fleeting. It is very difficult to keep a conscious image in mind<j <
for more than a moment. With rehearsal, we can refresh inner
speech to an extent, but even rehearsed words tend to fade and
satiate rather quickly (see below). Perceptual events can be
renewed in various ways, but we tend to stay with them only as
long as needed to achieve some specific goal, and no longer.
Conscious sensory memory, such as iconic or acoustical storage,
lasts only a few hundred milliseconds. One explanation of the
fleetingness of conscious events is that adaptive processes are
continually learning from conscious events, reducing Œtheir‹
alternatives, and nibbling away at the information provided. If
that is true, then most conscious events lose most of their
information value very quickly, as uncertainty is reduced by
adaptive mechanisms. The evanescence of conscious contents is
consistent with the notion of informativeness as a demand for
adaptation.
Finally, notice that habituated or automatic events can
become conscious again when their predictions are violated. If
they become unconscious due to redundancy, one way to make them
conscious again is to make them informative. And indeed, that is
what happens when the foot pedal of a bicycle falls off, and we
become conscious again of aspects of bicycle riding that were
largely automatic before. Steering the bicycle, balancing it and
the like, now become informative again, as they become conscious.
Again, consciousness is correlated with information content.
5.13 Redundancy Effects occur at all levels of processing.
We are now ready to discuss the strongest source of evidence
for the informativeness of conscious experience: the Redundancy
Effects. If we ask the average student of psychology whether
practice helps people to deal with new material, the student
would no doubt reply, "Yes, of course, repeating something helps
you to perceive and remember it." Rehearsing a telephone number
helps us to recall it, doesn't it? And if you do not understand a
sentence, reading it over again will surely give you more time to
think about it? There is however a fundamental class of cases in
which repetition harms rather than helps conscious access. When
we Œalready know‹ an event, repeating it further tends to harm our
ability to consciously perceive, understand, recall, and monitor
it. To say it slightly differently: whether repetition helps
conscious access or not depends upon the information value of the
event in question. As long as information increases with each
repetition, conscious availability will improve; but once the
stimulus is known, repeating it only creates more redundancy,
resulting in a Œloss‹ of conscious access. Table 5.13 presents this
evidence in summary form.
<h <
--------------------------------------------------------------------------------------------"
Table 5.13
-------------------------------------------------------------------------------------------?
Contrasts between informative and redundant phenomena (*).
ŒConscious phenomena‹ ŒComparable unconscious phenomena‹
Novel stimuli or concepts. Predictable repetitions
of habituated stimuli
Changes in habituated or concepts.
stimuli or concepts.
Novel actions. "Automatized" actions.
De-automatized actions.
-----------------------------------------------------------------
(*) The specific phenomena include: stimulus habituation with
repetition, stopped retinal images, automatization of practiced
skills, automatic visual images, semantic satiation,
inaccessability of predictable conceptual presuppositions,
habituation of the orienting response, and lack of conscious
access to routine contextual systems.
-----------------------------------------------------------------
Redundancy, the lack of information, is defined formally as
the presence of a signal after the uncertainty at the receiver is
already zero. Redundancy implies a perfect match of input with
expectation. The context of alternatives described in the section
above collapses to only a single representation, with one degree
of freedom, and maximum certainty. In the Piagetian dimension of
assimilation and accomodation, redundancy corresponds to the
extreme pole of assimilation.
Of course, most of the time when we pay attention to
something we do not wait for complete adaptation to occur. We are
satisfied to understand some idea or clarify some perceptual
event up to a point, and then go on to another. Redundancy
Effects present the extreme case of absolute adaptation, but most
of the time we are satisfied with Œrelative adaptation‹. Once the
conscious information is understood to some criterion, we go on.
We now turn to some specific examples.
<j <†
ŒPerceptual Redundancy Effects.‹
Redundancy Effects for the different senses are quite clear
in hearing, olfaction, taste, and touch; in all these senses a
repeated or lasting stimulus fades rapidly from consciousness. In
the case of hearing, Miller (1955) was able to show a rapid
decrement in the experienced intensity of even a single tone
after several hundred milliseconds. In olfaction, we can observe
every day that we lose track of a smell that may have been quite
noticeable when we first encountered it. Most people are quite
unaware of stable, surrounding odors, though these may be quite
obvious to a newcomer. The act of sniffing, which changes the
concentration of odor chemicals over the smell and taste
receptors, may serve to change the stimulus, to dishabituate
previously adapted receptors. Similarly, it is a common
experience that even the most delicious food becomes less
noticeable after the first mouthful. Gourmet cooking often
consists of selecting deliberate taste contrasts, to re-awaken
appreciation of otherwise habituated flavors. And the reader can
probably verify at this moment that a previously felt object ---
such as one's clothing --- has now faded from consciousness.
One might object that vision seems different. Consciousness
of visual stimuli does not seem to fade with repetitive
stimulation, even when we stare fixedly at an object. But this
neglects the existence of eye-movements. Our eyes are in
continual jumpy motion, both voluntary and involuntary, in
addition to head and body motion. As long as the eye keeps
scanning a differentiated visual field, new information will
enter the retina. Therefore there may always be some element of
uncertainty at the level of retinal receptors, as long as the
eyes keep moving. It is difficult to completely stop even large
eye-movements, and there is a continual tiny eye tremor
(physiological nystagmus) that cannot be stopped voluntarily at
all. Thus light under normal conditions, light patterns coming
into the eye are never wholly constant and predictable. However,
one can artificially defeat physiological nystagmus, and then
interesting things happen. For instance, Pritchard, Heron, & Hebb
(1960) mounted a tiny projector on a contact lens that was firmly
attached to the eye, so that the projected image moved along with
all eyecmovements. Under these conditions the visual image fades
in a few seconds. The method of "stabilized retinal images" shows
that even vision is subject to rapid habituation when it is not
protected by constant eye movements. Similarly, when people look
at a bright but featureless field (the Ganzfeld), they experience
"blank-outs" --- periods when visual perception seems to fade
altogether (Natsoulas, 19xx). It seems therefore that conscious
experience of highly predictable stimulation fades rapidly, in
all sensory modalities, without exception.
In nature, all the senses continually change their
relationship to the input: the nose sniffs the air, changing the
concentration of odors; the tongue tastes in an exploratory way,<j <
with the same effect; the hands and body explore by moving and
touching (haptic touch); in most mammals the ears can be pricked
up and oriented, and even humans tend to "cock" one ear to the
source of sound when listening carefully; and of course our eyes
are ceaselessly searching for significant information. Thus the
same physical event can enter a sensory system in many different
ways, so that habituation can be avoided. But in the laboratory
we can show that truly constant input fades rapidly from
consciousness.
Note again that conscious fading does not mean that the
event has disappeared from the nervous system. The Sokolov
argument (1.xx) shows that fading does not involve a fatiguing of
receptors, or anything else that is dysfunctional. Rather, the
fact that some stimulus fades from consciousness after repetition
is highly functional; it is a sign that learning has occurred.
ŒConceptual Redundancy Effects.‹
Habituation is not limited to perception: it may occur at
all levels of analysis. Fading of conscious access to abstract
concepts is shown by "semantic satiation." A word repeated over
and over again will soon seem different, somehow meaningless,
estranged from its previous familiarity, as if it were being
pronounced by some particularly impersonal robot (Amster, 1964).
Semantic satiation is similar to stimulus habituation, but it
seems to operate on the level of abstract meaning rather than
sensation. It suggests, therefore, that the informativeness
criterion does not just apply to perception and imagery, but to
conceptual thought as well.
There has been some controversy about the empirical validity
of semantic satiation (e.g. Esposito & Pelton, 1971), but the
evidence for conceptual redundancy is actually quite pervasive.
It is most common for experts in any discipline, who do not need
to communicate their expertise to novices, to find it difficult
to retrieve their knowledge explicitly (Anderson, 1983). This is
likely to happen even though the inaccessible knowledge continues
to inform their actions and experiences.
Research on Adaptation Level theory indicates that
conceptual events are evaluated in a context of alternatives that
is shaped by previous experience with the same events (Helson,
1964). Thus, one's judged happiness is strongly affected by
previous judgments of happiness. In general, when one achieves a
new level of desired functioning --- in getting a higher level of
income, a desired job, or a desired mate --- people report high
levels of happiness for some time. However, the new situation
rapidly becomes adapted to, so that now one evaluates events Œwith
respect to‹ the new Adaptation Level (AL). The reported level of<j <
happiness declines relative to one's expectations. In addition,
people frequently raise their sights again, so that they become
unhappy relative to their new desired state. Naturally, the
Adaptation Level at any particular time --- the level of one's
predictions about the world --- is not immediately consciously
available, though it is established by conscious experiences, and
it will become conscious again upon violation (5.xx). In general,
it may be said that those aspects of the world that we have
learned Œmost completely‹ tend to be Œthe least‹ conscious.
ŒRedundant goals also fade from consciousness, even though
they continue to operate as goals.‹
We can now make a direct connection between goals,
information, and consciousness, because redundant goals are lost
From consciousness as well (7.45). A graduate student may be ve
much aware in the beginning of graduate education of his or her
goal of obtaining an advanced degree. But this goal tends to fade
into the background as long as it remains highly predictable.
Everyday concerns have more to do with subgoals needed to carry
out the goal of gaining an advanced degree, than with this topTlevel goal. When the
student goes to find a book in the library,
it is not necessary to be reminded that this is done in pursuit
of the ultimate goal. A subgoal can even fail without bringing
to mind the highclevel goal: one may fail to find the right book
in the library, or one may even fail to pass an examination.
However, if all subgoals fail without alternatives, the top goal
also comes into question and must become conscious again. If
money for school runs out, if the student has a disabling
accident, etc., the topclevel goal of gaining an advanced degree
comes into consciousness, as alternatives to it need to be
examined. But as long as goals are predictable, they are as
unconscious as any other potentially conscious content.
In actual fact, the goal of gaining a graduate degree may be
a poor example, because it is a socially agreedcupon goal, one
that can be communicated to others for whom it is not redundant
but informative. Thus the goal of gaining a graduate degree is
fairly easy to access, even when it becomes routine, because it
often needs to be communicated. In the same way, teachers may be
able to access information that practitioners with the same
knowledge allow to become unconscious, because teachers must
always be ready to communicate the presupposed information to
students who do not share the presuppositions. From this point of
view, it seems likely that constant private goals may be much
more difficult to make conscious. Thus the goal of gaining the
respect and affection of others may become presupposed and
unconscious, even while we pursue its subgoals. Or the goal of
advancing one's social control, or to outshine competitors, may
become unconscious and still be pursued. From this point of view,
adaptation to goals may behave much like repression (7.xx) ccc<j <
that is to say, people will spontaneously disavow having such
goals, even though the unconscious routine goals will continue to
guide their actions.
ŒRedundancy Effects are not limited to conscious processes:
all neural structures adapt to predictable input.‹
Conscious experience is not the only thing that habituates.
Selective habituation to repeated input seems to be a universal
property of neural tissue (Kaidel, Kaidel & Weigand, 1972). Even
a single neuron will respond to electrical stimulation at a given
frequency only for a while; after that, it will cease responding
to the original frequency, but continue to respond to other
frequencies. Thus a pulse train of 300 Hz. will cause the neuron
to fire, until it habituates. After that point, it will no longer
fire to a 300 Hz stimulus, though it will continue to respond to
stimuli less than 280 Hz or more than 320 Hz (for example)
(Kaidel, Kaidel & Weigan, 1972). Neuronal habituation is
selective, just as habituation of the Orienting Response is
selective (1.xx). Further, Sokolov's arguments also seem to apply
at this level: fatigue canot be explain selective habituation,
because the neuron continues to be respond to nonchabituated
frequencies.
This kind of selective habituation can be observed at many
different levels of organization in the nervous system: in single
cells, in small assemblies of neurons, in larger nuclei and
pathways, in complete sensory and motor systems, and in the
nervous system as a whole (Tighe & Leaton, 1976). This point is
very important in our theoretical development because it suggests
that Œall‹ specialized processors attempt to adapt to (match)
input, and become quiescent when they have done so. But what
then is the difference between local neural habituation, and loss
of conscious access to some experience? Below, we will propose
that loss of conscious access is a global result of many cases of
local habituation by specialized processors (5.xx).
5.14 Adapted systems can provide new context for later
conscious experiences.
What happens after some conscious event is matched exactly?
We have just stated that the input does not disappear; what does
it do instead? Chapter 4 suggested that a major function of
consciousness is to evoke the proper context for later conscious
experience. We can now extend this point to the Œcreation‹ of new
contexts. One interesting possibility is that systems that adapt<j <
to conscious input create new contexts, which then begin to shape
future conscious events. Going back to the epigraph for this
chapter, we can now begin to explain James' observation that
"what we hear when thunder crashes is not thunder Œpure‹, but
thunder- breaking-upon-silence-and-contrasting-with-it." Those
systems that have adapted to the silence, and that therefore
predict continuing silence, must make major changes as a result
of the thunder clap. Further, they have become contextual with
respect to the experience of thunder: They notify us not only
that thunder has occurred, but that it was very different from
the foregoing level of sound.
We will consider two examples in detail. First, we consider
the act of driving a car, and next, the case of scientific
concept development.
ŒA perceptual-motor example: adapted conscious contents can
create a new context.‹
When we first learn to drive a car, we are very conscious of
the steering wheel, the transmission lever, the foot pedals, and
so on. But once having learned to drive, we minimize
consciousness of these things and become mainly concerned with
the road: with turns in the road, traffic to cope with, and
pedestrians to evade. The mechanics of driving becomes part of
the unconscious context within which we experience the road. But
even the road can be learned to the point of minimal conscious
involvement, if it is predictable enough: then we devote most of
our conscious capacity to thinking of different destinations, of
long-term goals, etc. The road has itself now become contextual.
The whole process is much like Alice falling through the Looking
Glass, entering a new reality, and forgetting for the time being
that this is not the only reality. Things that were previously
conscious become presupposed in the new reality. In fact, tools
and subgoals in general become contextual as they become
predictable and automatic.
Before learning to drive there are many things we can
consciously consider doing with a gear lever or a steering wheel.
We do not know exactly how much force is needed to turn the
wheel, or how to put the transmission lever in its various
positions. We don't have the "feel" of these actions. These are
all open choices --- constrained, of course, by their own
previous contexts. But there are many degrees of freedom in our
representation of these factors.
Of course the nervous system continues to represent and
process the foot-pressure on the accelerator pedal and the force
needed to turn the steering wheel, even after adaptation. These
factors do not disappear when they are lost from consciousness:<j <
They have simply become predictable through an adaptive learning
process. Learning or adaptation may in fact be defined as a
reduction of alternatives in a domain (5.14). When complexity has
been maximally reduced, we have learned successfully. Our general
claim in this chapter is that reducing alternatives to a single
one leads to a loss of consciousness of the source of stimulation
(section 5.xx). Indeed, the loss of consciousness that occurs
with habituation and automatization can be taken as a sign that
learning is complete.
But why, when the act of driving becomes automatic, do we
become conscious of the road? Presumably the road is much more
Œinformative‹ within our purposes than driving has become. Dodging
another car, turning a blind corner, braking for a pedestrian ---
these are much less predictable than the handling of the steering
wheel. These are now the differences that make a difference. But
once the road itself becomes routine and predictable, it too can
become context for other events, and so on indefinitely.
Notice that Œgoals‹ provide some of the constraints for the
conscious domain. Indeed, goals involve one kind of context of
alternatives (4.0, 6.0). Paying attention to choices at street
crossings determines whether we shall get to our destination (our
immediate goal), and driving safely determines whether we shall
survive (an enduring goal). Like other contexts, these goals are
not usually conscious when they shape action and experience. In
the act of dodging another car, we do not consciously remind
ourselves that we want to survive. We can interpret purposeful
actions as having a kind of informativeness, making a difference
within a goal context of alternatives. Goal contexts specify a
set of alternatives that our actions serve to select, so that our
purposeful actions are also informative at various levels of
significance. Not running over that pedestrian selects our goal
of not harming people, of avoiding trouble with the law, and of
getting where we are going with minimum effort. These are all
significant goals --- some are more significant than others, and
some are sub-goals for deeper goals. In all these cases we can
think of conscious events and voluntary actions as being
informative, in the sense of selecting alternatives within the
contextual goal hierarchy (5.xx; 7.0).
A contextual system does not have to remain unconscious for
a lifetime. It can be Œde‹contextualized to become once again a
domain of conscious experience. Suppose we are accustomed to
driving a car with power-steering, and one day the power-steering
fails. Suddenly the steering wheel becomes harder to turn,
especially at slow driving speeds. Our contextualized predictions
about the force needed to turn the wheel are violated; our
strategy for driving the car must change. Now the act of moving
the steering wheel becomes conscious again. Previously
predictable events decompose into several different alternatives,
as previously contextual processes become the objects of
conscious experience. We notice that we must keep the car moving,
in order to turn the wheel with less effort. This was not a<j <
consideration before, but it has now become relevant. Presumably
a global display of the newly conscious information helps to
bring such new considerations to bear upon our overall goal of
steering the car. This phenomenon of decontextualization of
previously unconscious tools and effectors is very general. When
we break an arm, this normally presupposed part of the bodily
context becomes an object of experience, one that now involves
conscious choices; the arm, too, has become decontextualized.
This view gives a new perspective on context (4.0): In a
sense, context consists of those things to which the nervous
system has Œalready‹ adapted; it is the ground against which new
information is defined. The same point can be made at the
conceptual level, as we see next.
ŒThe case of conceptual presuppositions: conscious contents
can turn into new contexts.‹
The development of conceptual presuppositions provides
another example of context-creation by adaptation. We have
previously made the case that all consciously accessible concepts
exist in a framework of prior unconscious presuppositions (4.x).
Without such presuppositions, the concepts themselves are
different. We now suggest that this presupposed knowledge is
simply the result of previously conscious concepts. When we first
encounter someone from a different culture, we become conscious
of many things that are normlly presupposed: the person may
speak from an uncomfortably close distance, he or she will have
different conceptual presuppositions, etc. If we live in the
foreign culture, all these prominent features disappear, and now
we are shocked when we discover our old culture again. Previously
presupposed ideas and actions now become conscious. As usual, it
is contrasts and transitions that bring out these points. Experts
have much less conscious access to such material than novices.
We can usefully refer to these phenomena in terms of
Œcontextualization‹ and Œdecontextualization‹. When we encounter some
new assumptions about reality, we often need to make them
explicit. That does not mean that we must define them verbally;
it may be good enough to contrast two different points of view, a
politically rightist vs. leftist viewpoint, for example. Once we
become familiar with the contrasts, they can become automatic and
contextualized, so that they will shape subsequent thought
without becoming conscious. If the new context is then violated,
however, it must become decontextualized; some contrast between
right-wing and left-wing politics, which was previously
unconscious becomes conscious again. (Decontextualization is
really the same as objectification, mentioned above.)
Take the process of constructing scientific reality, as in<j <
the discovery of the atom. In modern times, the first serious
proposals for the existence of atoms go back to George Dalton
(18xx - 18xx), who discovered that an electrical current will
decompose water into two parts hydrogen and one part oxygen.
Dalton proposed that tiny indivisible (atomic) particles of
hydrogen and oxygen must combine in a twoctocone ratio to form
water. But this hypothesis did not establish the reality of atoms
and molecules; it merely began a debate that continued for the
rest of the 19th century, with many facts pro and con being
offered by both sides. Some scientists refused to believe in
atoms, or only treated them as useful fictions, Œfacons the
parler‹, without any reality. They had many arguments in their
favor. Not all substances fell apart into elements with simple
ratios. The relationships between the supposed elements were
confusing, and could not be organized coherently until quite late
in the century. The reality of atoms was not universally
recognized until the Œvarious alternatives were shown to reduce to
essentially one interpretation.‹ This reduction in alternatives
culminated with Einstein's work early in this century (Kuhn,
1962). At that point, the reality of atoms became the conceptual
framework of a new world-view; no longer were atoms considered to
be merely convenient fictions. They were "real" objects.
In general, it appears that scientific constructs are not
considered "real" until other ways to interpret the evidence are
lost from sight. At that point the community stops arguing about
them, and begins to take the new construct for granted (Baars,
1986a). Indeed, the newly "real" objects can become the fulcrum
of novel explorations that now Œpresuppose‹ the existence of the
construct. Thus atoms have become part of the unquestioned
context within which modern physicists are exploring subnuclear
particles. For scientific constructs like atoms, it is not so
much that they disappear from consciousness of the scientists who
accept their reality. Rather, the construct is accepted when
alternatives are forgotten.
Thus in particle physics today no one challenges the reality
of atoms; to do that would undermine the task of exploring
subatomic and subnuclear particles. It would force researchers to
challenge the context within which protons and quarks are
defined. Challenging context is not impossible, of course.
Einstein's relativity theory decontextualized Newtonian
presuppositions about space and time, and quantum mechanics
decontextualized Einstein's assumptions about determinism (Kuhn,
1962). In both cases, physicists became conscious once more of
the alternatives to their presupposed reality. But it is very
difficult to decontextualize one's assumptions Œand also‹ at the
same time engage topics Œwithin‹ that set of assumptions. In
driving the car, one cannot be absorbed in moving the steering
wheel and successfully engage the road at the same time. In
general, a context must remain stable, presupposed, and largely
unconscious, in order for us to engage the objects that are
defined within it.
<j <† These parallels between a perceptual-motor task like driving
a car and the pursuit of scientific reality are quite intriguing.
They suggest that consciousness of the perceptual world and
consciousness of a conceptual reality like science may follow
similar laws. Notions like predictability and uncertainty,
informativeness and redundancy, context of alternatives, and
decontextualization may have very wide application.
----------------------------------------"
Insert Figure 5.14 about here.
----------------------------------------"
The thrust of this section has been that human beings Œadapt‹
to information at all levels: perceptually, in imagination,
skills, concepts, and goals. As they adapt, they lose conscious
access. The next section argues that people also Œseek‹ informative
input, so that there are actually two countervailing tendencies.
5.2 Human beings also Œseek‹ information at many levels.
So far we have discussed ways in which the nervous system
adapts to some conscious event, thereby Œreducing‹ conscious access
to it. But the opposite process occurs as well: there is
extensive evidence that people seek out novel and informative
conscious contents. We do not wait for the perceptual world to
fade. We always go on to seek new and interesting things. In sum,
there seem to be two tendencies: a tendency to reduce conscious
access by adaptation, and a countervailing tendency to increase
conscious access by searching for informative stimulation. These
two tendencies may alternate, so that we seek conscious
information, adapt to it, seek a new source of information, adapt
to that, etc. (Figure 5.2). The process may approach a self-
regulating homeostasis that tends toward optimal information
flow. In this section we explore the Œsearch‹ for information at
different levels of conscious access: in perception, in
conceptual processes, and in the domain of goals, where the
search for information helps to define the significance of
conscious input.
c------------------------------------
Insert Figure 5.2 about here.
--------------------------------------
<j <†5.21 Perceptual systems aim for informative parts of the field.
In nature, all of an animal's senses work together in
active, coherent exploration. Given a surprising noise, a dog's
ears will prick up, it will look toward the sound, its pupils
dilate, lungs begin sniffing the air, nostrils flare to allow
better smelling; the animal will even taste the in-breathed air
as it flows over the tongue. If the sound is interesting the dog
will move toward it, constantly sniffing, looking, listening, and
licking anything of interest. It is actively searching for
information --- for signals that make a difference in the search
for food, for social and sexual partners, for dangers to avoid,
and often for plain novelty.
In the laboratory, by contrast, we usually examine only one
perceptual system at a time; but the same overwhelming preference
for information emerges there. There is extensive evidence that
eye-movements focus on the most informative parts of a scene
(Yarbus, 1967; Kahneman, 1973). Given a choice between
predictable and unpredictable figures, people choose those that
are moderately unpredictable: those with enough information to be
interesting, but not so much as to be confusing or overwhelming
with novelty (Garner, 1974). And it is well-established that
animals and people will work for informative stimulation, without
food or any other reward (e.g. Harlow, 19xx).
As we see next, the same restless search for information
characterizes conceptual processes, those that are abstract and
not directly reducible to perception or imagery.
5.22 Conceptual processes aim for informative points.
"Be informative" is a cardinal rule of normal discourse
(Clark & Clark, 1977). In fact, violations of this rule are quite
strange. In fact, violations of this rule are quite strange. When
the same conceptual message is repeated over and over again, we
tend to turn away to other, more interesting and informative
material. If we nevertheless Œtry‹ to pay attention to the same
redundant material, we find that doing this is quite effortful,
and ultimately impossible (Mackworth, 1970). People do not ask
questions about the things they already know. We always speak
with a point of information in mind, either for the speaker or
the listener. All these facts suggest that people Œseek‹ conceptual
information in the sense described above.
But what about apparent exceptions, such as repeated<j <
insistent demands for help, or a child's pursuit of some desire?
Surely messages like this can be repeated hundreds of times
without adding new information. What about obsessive thoughts,
which may recur thousands of times? All of these cases can be
reconciled with the idea that people search for novel
information, if we interpret them within a goal context. Goal
contexts are much more lasting and invariable than perceptual
contexts, and the same perceptual message --- can I have that
toy? --- may be repeated over and over again without losing its
informativeness in the goal context (5.4x).
We can again describe these phenomena in terms of Figure
5.2, with the difference that conceptual processes involve a more
abstract level of representation. Clearly the search for
information can operate at many levels or representation, just as
adaptation to information occurs at all levels of representation.
5.23 Goal contexts define significant information
to attend to.
Much of normal conscious thought is devoted to goals and the
means to achieve them. Thought-sampling studies show that college
students devote about 40% of their time to clearly goal-related
thoughts and images (Klinger, 1971; Singer, 1984). But even
thoughts that Œseem‹ purposeless may be driven by goals that are
not currently conscious (Pope & Singer, 1978; Horowitz, 1975 ab;
see Chapter 6). Even daydreaming may serve specific goals.
In this section we cite evidence that goal contexts define
"significant" information; i.e., that people are highly sensitive
to signals that reduce uncertainty within goal-defined contexts.
Those signals become the objects of attention: we seek to make
them conscious. As in the example of a child demanding a new,
fascinating toy, this point implies that input may be repetitive
at other levels, but informative in a goal context. We are
willing to seek out information that is redundant at the
perceptually and conceptually, as long as it is significant
within a dominant goal context.
Goals are more important than arbitrary stimuli or concepts;
indeed, psychological significance is defined by goals. They
provide the higher-level context of information for any animal.
Eating and drinking, avoiding danger and discomfort, competing
and cooperating with other animals --- all these activities are
defined by goals. In that sense, goal-related input is inherently
informative (6.0, 7.0). We can see this in the case of stimulus
habituation: when a stimulus signals food to a hungry animal, it
can be repeated many times before the orienting response
habituates. But when the animal is sated, repetition of the same<j <
signal causes rapid habituation of orienting (Sokolov, 1963).
Conditioning generally involves reinforcers that tap into
biologically significant goals --- food, drink, avoidance of
pain, and the like. And classical conditioning is believed to
depend heavily upon the amount of information given by the
conditional stimulus about a significant (goal-related)
unconditional stimulus (Rescorla & Wagner, 1972).
It is obvious that people scan the world for significant
information, and make it conscious when possible. Attentional
control fades quite quickly in spite of our best efforts, but it
can be revived by creating pay-offs for successes and failures;
that is, when we create a direct connection between conscious
vigilance and a currently significant goal (Mackworth, 1970).
When there is a monetary reward or penalty, when survival depends
upon vigilance --- as in the case of a war-time submarine sonar
operator --- then we can to some extent overcome the tendency of
redundant stimuli to fade from consciousness. Thus the claim that
goal contexts define significance, and that we actively search
for such significance, seems well-supported. This connection is
modeled here by showing that we can recruit a goal context to
help maintain relevant information on the GW.
-----------------------------------------------------------------
FIGURE 5.23: information may be redundant perceptually and
conceptually but still trigger goal contexts.
----------------------------------------------------------------
5.24 Summary: Seeking vs. adapting to information.
This chapter has noted two countervailing tendencies: the
search for information and adaptation to information. The first
leads to more conscious access, and the second reduces conscious
access: Obviously our model should reflect both. So far, the
picture seems quite consistent across the board. Perceptual
systems are highly sensitive to information rather than physical
energy. Conceptual and goal processes can similarly be viewed as
sensitive to information --- to distinctions that make a
difference, and that trigger adaptive processes. And there is a
special relationship between information and consciousness, as
shown by the fact that redundant stimuli fade from consciousness.
The exceptions to this rule are discussed in section 5.4 below.
Note again that in ordinary life, adaptation does not have to be
complete: when routine events become partially redundant, other,
more informative events demand our attention.
5.3 Model 3: Interpreting informativeness in the theory.
We can now begin to describe this pattern of evidence in
terms of GW theory. Figure 5.3 describes the facts, showing how
input can serve to select one of several alternatives, a-e, that
are defined within a stable context. As choice "b" is selected
over and over again, the context begins to predict it routinely,
until finally the presentation of stimulus "b" no longer requires
conscious involvement. At this point a new unconscious connection
may be established between the input and its mental
representation, or possibly there is a momentary global display
for shorter and shorter periods of time, until it becomes very
difficult to report (2.xx). (The dashed arrows indicate that both
alternatives may be possible.) The more predictable "b" becomes,
the most redundant it is with respect to its mental
representation, and the less it is conscious.
cccccccccccccccc
Insert Figure 5.3 about here.
cccccccccccccccc
5.31 How does the system know what is informative?
How does the GW system determine whether some input is
informative? Sometimes we are told that some source of conscious
information is important: that is, we accept the goal to pay
attention to it (see Chapter 8). Indeed, that is what we did with
the Dalmatian example --- the reader was simply asked to pay
attention to the demonstration. Thus sometimes what we make
conscious is under the control of goals. However, even when we
try to pay attention to a boring and repetitive stimulus, it
becomes difficult to do so after a short while. Other thoughts
come to mind, and the input fades in spite of our best efforts to
pay attention to it. Thus there must be some way in which the
system can determine how informative the input is independent of
goal-controlled attention.
The most plausible supposition is that the audience decides.
Specialized processors that are interested in the global message
may feed back their interest, and do so until all the usable
information is absorbed, or until some other conscious content
becomes more informative. We have previously called this the
"Nielsen ratings of the mind," by analogy to the continuous
assessment of the popularity of different television programs in<j <
the United States. In Model 3 we show this as a feedback loop
coming from the receiving processors to the global message.
Presumably we lose conscious access to a repeated stimulus if the
receiving processors stop feeding back their interest. Figure
5.31 shows this kind of feedback loop.
--------------------------------------
Insert Figure 5.31 about here.
--------------------------------------
Notice that Model 3 is quite consistent with the
physiological evidence we discussed in Chapter 3 (Model 1A).
There are indeed feedback loops coming from the cortex and
elsewhere to the ERTAS system. It is also consistent with the
fact that all neural structures, down to single neurons,
habituate in a stimulus-specific way (5.xx). Specialized
processors receive global information, and, as they adapt to it,
they no longer respond with feedback requesting more of the
global message.
This is really a kind of coalition formation between the
receiving processors and the processors that support the message.
It is as if television sets have feedback monitors that let the
broadcasting station know how many people are watching. When the
program is popular, the audience supports the input processors
--- the actors, writers, and producers of the show. There is a
coalition in support of the conscious content. But as the
audience adapts to the broadcast, it become predictable and
uninformative, so that fewer and fewer audience members continue
to watch. The coalition breaks up, and may re-form around another
global message.
In summary, conscious experience suggests that the receiving
systems are feeding back their interest in the conscious global
message (5.0). Another way of stating this is to say that any
conscious message must be Œglobally informative‹.
5.32 It takes time for GW input to become conscious.
Creating a coalition of global senders and receivers
presumably takes time. A short tentative global message may first
be broadcast. Some receivers may request more of it, resulting in
a longer message, which gathers more support, and so on, in a
"snowballing" fashion, until finally the global message becomes
available long enough to report as a conscious event (3.xx;
2.xx). We have previously referred to this notion as the
Momentary Access hypothesis (x.xx). It is consistent with Libet's
(1979) finding that skin stimulation of moderate intensity may
take as long as .5 seconds to become conscious, even though
cortical activity from the stimulus can be recorded long before<j <
that time.
5.33 When one conscious event becomes redundant, the next
most informative input becomes conscious.
The model does not imply that all conscious events are
completely adapted to until they are utterly redundant. Most of
the time when we read or hear a sentence we do not wait for
complete adaptation to take place --- we only need to adapt to a
single aspect of the input. The reader of this book is not going
to repeat each word or sentence over and over again to the point
of semantic satiation --- it is enough to simply wait for the
"click of comprehension." Instead, the model suggests there is
relative adaptation: We adapt to a conscious event to some point,
perhaps until we feel that we have comprehended it. After that,
other potential conscious contents may be scanned (8.xx), to see
if there is one with greater significance, or with greater
ability to recruit a coalition of receivers to support its
dominance.
5.34 Informativeness and objectification, redundancy and
contextualization.
We have suggested that during the information stage of
adaptation, there are choices in the stimulus, either implicit or
explicit. To see even a simple black line on white paper means
that there is an implicit figure-ground comparison; the
brightness of the line and paper are implicitly compared to
previous visual intensities, as well as to adjacent contrasting
areas; and so on. There is no conscious content without implicit
comparison. This point has an interesting bearing on the general
issue of contextualization vs. objectification. Conscious
contents seem always to be object-like. Even abstract entities
and processes tend to be reified and treated as objects --- we
speak of "mathematics," "democracy," and "process," as if they
are objects like chairs, tables, and pencils. Of course when
these ideas become thoroughly predictable, they become habituated
and automatic, and fade from consciousness, though even then they
can constrain future conscious experience. That is to say, they
have become context, by definition of that term. We can call this
process Œcontextualization‹. The reverse occurs in the case of
violated presuppositions discussed above. Presupposed contextual
constraints can become conscious when they are violated, and
hence they become object-like (decontextualized). They are taken
From the status of context and objectified --- turned into
object of experience and thought. The notions of
contextualization and objectification have wide-ranging
consequences (see Chapter 9).<j <†
5.4 When repeated experiences do not fade: Is informativeness a
necessary condition for conscious experience?
The claim that informativeness is necessary for a conscious
event depends heavily upon the Redundancy Effects discussed above
--- those cases where input is repeated over and over again and
consequently disappears from consciousness. These effects are
pervasive: they exist in all senses, in mental imagery, in motor
skills, and apparently in conceptual processes as well. However,
if there are clear exceptions to the Redundancy Effects, the
hypothesis of a necessary connection between consciousness and
information cannot be maintained. The existence of genuine
counter-examples would destroy the "necessary condition" claim.
In this section we discuss some apparent counter-examples and
show that these can generally be handled in a broad information-
theoretic framework. The "necessary condition" claim, we
conclude, seems quite defensible.
5.41 The apparent implausibility of Redundancy Effects in
everyday life.
On the face of it there is something implausible about the
idea that Œall‹ conscious experiences fade when they become
automatically predictable. If that were true, how could we
experience the same road to work every day of our lives? Or the
same kitchen table, the same bedroom, the same faces of friends
and family? Is there a Œprima facie‹ absurdity in the
informativeness claim? Some Redundancy Effects, like the stopped
retinal images discussed above, occur only under laboratory
conditions. Perhaps the laboratory creates artificial
circumstances that do not represent the underlying reality?
Some of the counterexamples pose genuine challenges, and
others do not. For instance, Neisser trained people to search for
a single word in a list of words, or a face in a photograph of a
crowd (see Neisser, 1967). People can learn to do this very well,
so that the target face seems to "pop out" of the crowd quite
automatically. In these experiments, automaticity in the task
seems to Œlead‹ to conscious access, rather than causing it to
fade. But this is a false counterargument. If the task practiced
involves conscious access, then of course practice in this task
should lead to more efficient conscious access. What becomes
automatic in the Neisser studies is the act of attending, the act
of making things conscious, as opposed to the object that becomes
conscious. The particular face in the crowd, or the fact that<j <
this face is to be found in this particular place, is quite new
and informative (see 8.00). If this is true, the act of attending
should fade from consciousness even if the target does not, and
the Neisser studies do not provide a true counter-argument to the
claim that repeated conscious contents fade with redundancy.
Some other counterarguments are more difficult to handle.
There are in fact clear cases where habituation of awareness does
not hold, where we continue to be conscious of repeated
experiences. Pavlov observed "spontaneous dishabituation" among
dogs exposed to repeated sounds, and such things as chronic pain
never permanently fade from the consciousness of its victims.
Even stopped retinal images do not fade permanently; they tend to
reappear in a transformed way, so that the word "BEER" will fade
and reappear as "PEER," "PEEP," "REFP," etc. These cases may
represent true counter-examples. We will argue below that we can
retain the informativeness hypothesis in spite of these
counterarguments, if we take into account the fact that repeated
material Œcan remain informative if its context of interpretation
changes‹. This is true of the formal definition of information
(Shannon & Weaver, 1949), which permits a repeated message to
continue to yield information if the context of choices within
which the signal is received is altered. Information is a matter
of the relationship of a message to its context, not of either
message or context alone. Thus, in cases where a repeated event
does not fade, we can ask whether its context of interpretation
has changed.
We now explore these issues in some detail.
ŒSome apparent counter-examples‹.
Consciousness to repeated events is Œnot‹ lost in the
following cases:
(1) Variability. The event seems to be repeated, but in fact
there is variability in the input. Visual information provides a
good example. We would expect that looking at a clock would cause
the stimulus to fade from consciousness if the informativeness
hypothesis is correct. In fact, the clock seems to stay in
consciousness; but of course, this is only because we cannot
control involuntary eye movements, especially physiological
nystagmus. We only seem to be staring fixedly at the clock --- in
fact, the input is variable. Similarly, automaticity of skill
learning does not occur if the skill is in fact variable
(Shiffrin & Schneider, 1977).
(2) Learning. Fading does not occur if the repeated<j <
stimulus is incompletely learned, so that it is better understood
with each repetition, and hence has not become truly redundant.
This is of course the classical learning curve, the relationship
between practice and learning which is so much better known than
the equally common Redundancy Effects.
(3) Relative adaptation. Fading does not occur if we do not
repeat the event to the point of complete adaptation. We stay
conscious of an event if we move on to another one as soon as
enough adaptation has occurred. We try to listen to a repeating
owrd, but in fact, a thought, a feeling, or an image begins to
come to mind, even before the target stimulus fades. Most of the
time we are satisfied with relative adaptation, as noted above.
(4) Ambiguity. Redundancy does not occur as quickly if the
stimulus is ambiguous and can be reinterpreted, so that it is
consciously experienced to be different.
We have already discussed the prevalence of ambiguity in the
world (2.xx). Language is rife with ambiguity; the social world
is profoundly ambiguous; the future is unknown and ambiguous;
bodily arousal can often be interpreted in more than one way;
conceptual ambiguity is prevalent, even in science.
Good naturalistic examples of spontaneous reinterpretation
of ambiguous events can be found in art, literature, music,
mathematics and science. Any piece of polyphonic music shows
figure-ground ambiguity. If we pay attention to one melodic line,
the others will fade into the "ground" of our perceptual
experience. In this sense all polyphonic music can be experienced
in many different ways. Further, even a single melodic line can
be reinterpreted, because we are continually generating
expectations about the next few notes before we hear them; if the
composer is clever he will occasionally surprise us with an
unexpected melodic turn. Truly great composers continually
surprise and please us by the interplay of the predictable and
unpredictable in the fate of an ongoing melody.
Musicians often find new and different sources of pleasure
in the same composition, even when it is played hundreds of times
over a period of years. Rather than fading from consciousness,
the music is continually reinterpreted. This is true for other
art forms as well. Serious works of art cannot be understood
completely on first exposure. They require a many-leveled process
of reinterpretation before we fully appreciate them.
Reinterpretation can happen spontaneously, simply by allowing
oneself to be conscious of the work, or under guided voluntary
control.
(5) Significance and purpose can keep repeated information
in awareness. Fading does not occur even with a repeated stimulus<j <
if it has significance beyond itself which has not been adapted
to. Redundancy may be avoided if there is an ongoing Œgoal‹ in
attending to the repeated stimulus, especially if the observer
receives feedback on the success or failure of his goal.
Presumably the more important the goal, the more we resist
redundancy, because the stimulus, which may be redundant
perceptually, conveys deep significance.
Suppose two people are driving in a car to a new
destination, but only the driver is involved in finding the way
--- the passenger is just enjoying the ride. When they go to the
same place a week later, which one is likely to remember the way
better? Common observation suggests that the driver will, even if
the passenger experienced the same physical flow of stimulation.
The difference is that the driver engaged the world in a
purposeful way, wondering whether to turn here or there, noting
distinctive landmarks at critical choice-points, and the like.
The driver's conscious experience was guided by a set of
purposes, while the passenger's experience was relatively
purposeless.
If people have different experiences of a single event when
they are guided by different purposes, their memories should also
be different. Thus Pichert and Anderson (1977) presented the same
story about two boys playing in a house to two groups of
subjects. The first group was told to take the perspective of a
home buyer, while the second group assumed the viewpoint of a
burglar "casing" the house. Different facts were recalled by the
two groups. The "home buyers" were more likely to remember a
leaking roof, while the "burglars" were more likely to remember
the location of the color television (Bransford, 1979). This is
consistent with the view that different purposes yield different
experiences --- or, as we will argue below, inner contextual
changes can create new experiences of the same event.
One effect of purpose is voluntary release from habituation
--c one aspect of voluntary attention (8.00). We can voluntarily
make unconscious habituated stimuli conscious again. Simply by
choosing to pay attention, the reader can again become conscious
of the feeling of the chair, of the background noise, of the
quality of the ambient light, and even of semantic
presuppositions. We will not model this volitional phenomenon
until Chapter 8, where we discuss voluntary attention. Note
however that attempts at voluntary control shift the internal
context of the signal. Thus this example seems to fit the claim
(detailed below) that an internal shift of context can take place
even if the physical input is repetitive, resulting in a new
conscious experience.
(6) Contextual shifts. Repetition sometimes leads to
spontaneous perceptual transformations. When we listen passively
to a repeated word, we soon begin hearing different words
(Warren, 1961, 1968). Within a minute or so a word like "break"<j <
will be begin to be heard as "rake, wake, wait, rape, ape, ate,
ache, ..." etc. This remarkable Verbal Transformation Effect is
different from semantic satiation (described above) because here
the subject is not saying the word, but merely listening to it.
A very similar phenomenon is observed with stopped retinal
images. In this case the transformations change according to the
visual properties of letters rather than following phonemic or
sound patterns (Pritchard, Heron, & Hebb, 1960). Thus BEER will
turn to BEEP, because the P and R are visually similar, while in
the auditory case "break" may change to "wake" because the /r/
and /w/ are phonetically similar (Lackner & Goldstein, 1975).
Because our knowledge of the acoustic properties of speech
has improved dramatically over the past twenty years, it has been
possible to examine the Verbal Transformation process in detail.
It is well-established that sounds like /ba/ and /pa/ differ in
only one articulatory property. In /ba/ the vocal chords begin to
vibrate a few tens of milliseconds before the lips open, while in
the case of /pa/ the lips open a short time before the start of
voicing (Liberman, Cooper, Shankweiler & StuddertcKennedy, 1967).
Using computer generated speech one can systematically vary the
"voice onset time" (the time difference between voicing and
opening the lips), and locate the exact boundary between the
perceived /pa/ and /ba/. Now we can examine the effects of
selective habituation. If /pa/ is repeated over and over again,
the boundary will shift in the direction of opening before
voicing; if /ba/ is repeated over and over again, the reverse
occurs (Goldstein & Lackner, 1975; Lackner & Goldstein, 1975).
This effect has been shown with natural as well as
computer-generated speech. The implication is that if one day we
heard all /p/'s and no /b/'s, our perception of these sounds
would be grossly distorted, because the phonetic boundaries would
shift. But the distortion would go in the right direction. If we
heard all /p/'s, the /b/ category would expand, so that more and
more cases would be interpreted as /b/'s. Thus the phonetic
system acts to regulate itself, and to maintain a relatively
constant number of /b/'s and /p/'s. In addition, the actual
frequency of these "opponent" phonemes in the language is roughly
the same ccc we normally hear rougly equal numbers of /b/'s and
/p/'s ccc so that the boundary stays at the same voice-onset
time.
Thus the context of the information can vary, but the system
is designed to keep it reasonably stable under most
circumstances. Generally speaking, the stability of our
perceptual contexts depend upon the existence of variation in the
perceptual contents. The function of the distribution of /p/'s
and /b/'s in a language may be to create enough variability to
maintain the categorical boundary. This is similar to the case of
physiological nystagmus, the function of which may be to avoid
excessive redundancy of input. A similar argument also applies to
opponent processes in other senses, such as color perception
(Gregory, 1966). Thus context can change, especially if we are<j <
exposed to only one end of a continuum of variation.
This conclusion also comes from the Adaptation Level (AL)
Theory of Helson (Helson, 1964). Our ability to specify the
"expected" stimulus value in perception or judgment depends
largely on our experience of the extremes in the same dimension.
We will judge criminality with less severity if we are routinely
exposed to rapes and murders (if only on television). As a
result, our judgment of criminal severity changes. Political
radicalization may work through the same mechanism; the more we
become used to an extreme belief, the less extreme it seems.
Again, our experience of an event depends upon our previous
related experiences, even though these are not conscious at the
time. Again, the context changes when we are exposed repeatedly
to one end of the continuum of variation.
Now we can go back to our original question. Why are there
cases of repetition that do not result in conscious fading? One
plausible suggestion is that the context has changed. When the
same word is repeated over and over again, or when someone is
exposed to a stopped retinal image of a word, we observe fading,
but also transformation. The fading can be explained as a
Redundancy Effect, but existence of transformations requires
another explanation. The hypothesis is that as one extreme value
of an opponent process is repeated over and over again, the
context of interpretation shifts. This is clearly the case for
Verbal Transformation, and it is at least plausible for the
stopped retinal images in the visual system. Of course the same
physical signal in a different context creates different
information (Shannon & Weaver, 1949). In this way we can provide
a satisfying account for these interesting counter-arguments to
the informativeness hypothesis.
In sum: Repeated signals may not fade from consciousness if
they are incompletely known, so that each repetition allows more
information to be absorbed; if the signals are variable; if they
are ambiguous, so that they can be interpreted in different ways;
if they serve a larger purpose which is not redundant; or if the
context drifts, so that the same signal remains informative.
It seems therefore as if we can explain the apparent
counterarguments to the "informativeness criterion" for conscious
experience. Of course this question deserves much more testing.
5.43 Summary.
We have suggested that all conscious experience must be
informative --- that true redundancy leads to a loss of
consciousness of a message. The evidence for this is quite
pervasive, ranging from repeated stimuli in all sensory<j <
modalities, to repeated visual images, automatized skills,
semantic satiation, and even stable conceptual presuppositions.
There are counter-arguments which seem compelling at first, but
which are less so upon further examination. Although more
research is needed on these questions, the position that
informativeness is a necessary condition for conscious experience
seems to be quite defensible.
5.5 Implications for learning.
If consciousness always involves adaptation, there should be
an intimate connection between conscious experience and all kinds
of adaptive processes, including comprehension, learning, and
problem-solving. We now explore these implications for learning.
5.51 Conscious experiences trigger widespread adaptation and
learning.
Learning of all kinds is surely the most obvious adaptive
mental process in which people engage. To learn something
deliberately, we typically act to become conscious of the
material to be learned. But most details of learning are
unconscious.
Information and learning are closely related. The most
widely accepted model of classical conditioning is defined in
terms of informative features of the conditioned stimulus
(Rescorla & Wagner, 1972). Recent "connectionist" models of human
learning also rely on mathematical rules that maximize the amount
of information given by one event about another (Gluck & Bower,
1986; Sutton & Barto, 1981). These models do not have an
explicit role for conscious experience. However, they may be
moving in that direction.
From a theoretical point of view, we expect consciousness to
be involved in learning of novel events, or novel connections
between known events. The rationale is of course that novel
connections require unpredictable interactions between
specialized processors. Hence global communication from "any"
specialist to "any other" is necessary (2.xx). Widespread
broadcasting serves to make this any-any connection.
<j <† What is the evidence for this claim? Perhaps the most
obvious is the radical simplicity of the act of learning. To
learn Œanything‹ new we merely pay attention to it. Learning occurs
"magically" --- we merely allow ourselves to interact consciously
with algebra, with language, or with a perceptual puzzle like the
Dalmatian (5.xx), and somehow, without detailed conscious
intervention, we acquire the relevant knowledge and skill. But of
course we know that learning cannot be a simple, unitary process
in its details. The Dalmatian requires subtle and sophisticated
visual and spatial analysis; language requires highly specialized
analysis of sound and syntax; indeed all forms of learning
involve specialized components, sources of knowledge, and
acquisition strategies. Of course these specifics of learning are
unconscious when they operate most effectively.
The key step in deliberate learning is to become conscious of
what is to be learned. Doing this is sufficient to learn, as
shown by many studies of recognition memory. In general, if
people are just made to pay attention to some material as an
incidental task, recognition memory for the material will be
quite good, even a week later, provided that the material is
distinctive enough not be confused with very similar material
(Bransford, 1979). Thus consciousness seems to lead to learning.
Whether consciousness is a Œnecessary condition‹ for learning is a
more difficult question, discussed below (5.xx).
Finally, we are driven by our theoretical position to a
rather radical position about most learning. Very often, the end
product of learning is a Œchange in the context of experience‹; but
we know that a change in context in its turn alters subsequent
experience. It follows that learning alters the conscious
experience of the learned material. Evidence for this position
seems strong for perceptual learning, knowledge acquisition,
skill learning, immediate memory, episodic memory, and rule
learning. It may be more debatable for associative learning. We
explore this claim next.
5.52 Learning alters the experience of the material learned.
If it is true that learning involves the generation of new
contexts, and if contexts shape and bound new conscious
experiences, it follows that Œwe experience the same materials in
a different way after learning.‹ Is there evidence for this
implication? Certainly we talk about algebra as "the same thing"
before and after we learning it, just as we talk about the
Dalmatian demonstration (above) as the same "thing" before and
after comprehension. But both algebra and the Dalmatian are
experienced differently after learning. Perceptual learning<j <
certainly changes the experience of the stimulus. Children are
thought to experience the perceptual world differently after
acquiring object permanence, for example (Piaget, 1952). Native
speakers of a language can often discriminate phonetic
distinctions which foreigners cannot hear: Most English speakers
simply cannot hear the Chinese tonal system. Even in learning to
comprehend a puzzling sentence, there is a change in experience
(4.xx back-up sentence). In the "I scream/ice cream" example of
Chapter 2, the perceived word boundaries switch back and forth;
and indeed, one of the great difficulties in learning foreign
language is in learning to perceive the word boundaries.
Similarly, conceptual learning --- of the sort that a student of
science does in learning physics --- clearly involves a change in
perspective and insight into the field. The announcement that a
new subcnuclear particle has been discovered must lead to a
different experience of comprehension for an advanced listener
than for a novice.
What about associative learning? When we need to discover
the connection between two known stimuli, or between a known
stimulus and a known response, is there a change in conscious
experience? This is not so clear. Perhaps the strongest evidence
in favor of a change in experience comes from a series of
brilliant studies by Dawson and Furedy (1976). These researchers
showed that human GSR conditioning did not occur if subjects
misinterpreted the relationship between the conditioning stimuli.
In standard GSR conditioning a stimulus is given, such as a tone,
and followed by a shock which elicits a change in skin
conductivity (GSR). Dawson and Furedy provided this stimulus
situation, and normal conditioning occurred. But now they changed
the subject's mental set about the stimulus sequence. Subjects
were told that the task was to detect a tone in noise, and that
the function of the shock was to mark the boundaries of the
trials. (Experimental subjects will believe almost anything.)
That is, they took the stimulus conditions (tone-shock, tone-
shock, tone-shock) and made the subjects think of them in the
opposite way (shock-tone, shock-tone, shock-tone). Under these
circumstances, the tone no longer serves as a signal for the
shock. And indeed, even though the stimulus conditions were
unchanged, conditioning failed to take place.
What does this mean for the question of changing experience?
We still do not know whether associative learning changes the
experience of the learned connection. However, the Dawson and
Griggs studies show that Œif‹ we experience stimuli such as way
that the tone does not seem to signal the coming of the shock,
learning will not occur. It may be therefore that the distinction
between associative learning and knowledge acquisition is a false
distinction: all learning takes places within a knowledge context
that defines the relationships between the stimuli. If that is
true, then it seems likely that learning changes this knowledge
context even in the case of associative learning.
This somewhat radical hypothesis about learning has a<j <
perplexing implication. Namely, if we experience an event
differently after learning, why do we still think it is the same
event? That is, how do we maintain Œevent identity‹ before and
after learning? This is a profound and difficult question, which
was raised by William James (of course) (1890). It is also the
question raised by Kuhn (1962) about scientific constructs after
a paradigm shift. Indeed, scientific constructs like gravity and
light are quite different in Relativity Theory compared to
Newtonian physics: yet they are called by the same names, and
they are naively believed to be the same things. Many of
physical observations relevant to light and gravity are
unchanged, of course, but not all (Kuhn, 1962). And some new
relationships are added with the coming of Relativity Theory: the
bending of light by gravity, for example. Nonetheless, "construct
identity" is maintained, at least in the sense that many
physicists Œbelieve‹ that in the Einsteinian framework they are
simply understanding "the same thing" in a deeper way. The
general implication is that event identity is a function not only
of the observations in question, but of the entire knowledge
context in which it is defined. This is Œnot‹ just true in physics,
but in perception, in conceptual learning, and perhaps in
learning in general.
ŒThe developmental role of forgotten conscious choice-points.‹
There is another interesting implication of the hypothesis
that learning changes the experience of the learned material. Œ‹Any
developmental process must involve choice-points between
different potential paths. We may choose to learn to play piano
at the age of six; if not, we are unlikely to become concert
pianists. We may choose to distrust certain people at an early
age as a result of traumatic experiences, and thus avoid finding
out that our distrust is unjustified. And so on. At the moment of
choice, we may be quite conscious of the alternatives; once
having chosen, we enter a new context that is created by the
choice, and within which the choice is often not even defined.
Once having learned algebra, it is extremely difficult to
re-experience the confusion that was once so common about
meaningless algebraic symbols on a page. Thus we often cannot
make previous choice-points conscious once we have entered the
new context created by those choice-points. We are utterly at the
mercy of our previous choices, and we cannot undo them. This
suggests that learning, and its consequent alteration in
experience, is never fully reversible (viz., Mandler, 198x). This
is a point with major consequences for developmental psychology.
<j <†
5.53 Is consciousness necessary for learning?
The fact that learning begins with a conscious experience
is known to every parent and teacher who has ever tried to reach
distractable children. In daily life this is what the term
"attention" means: it involves an attempt to control what shall
become conscious (see Chapter 8). In the psychology laboratory we
always call the attention of subjects to whatever is to be
learned. But somehow the salience of this plain everyday fact has
escaped the notice of many researchers, in part because it has
been superseded by a controversy: that is the question whether
consciousness is a Œnecessary condition‹ for learning (e.g.
Erikson, 1962; Dixon, 1971, 198x; Marcel, 1983 ab; Holender,
1986). This controversy has been difficult to resolve
conclusively, in good part because it raises the difficult
question of defining empirically the "zero point" about
consciousness. We have previously remarked on this difficulty,
and on the importance of developing a theoretical approach that
does not require a solution to this extremely problematic
question (1.xx). Unfortunately in the case of learning, most
discussion of the role of consciousness seems to be assimilated
to the "necessary condition" question. But even if conscious
experience were not a necessary condition but only a helpful
adjunct to the learning process, it would be difficult to doubt
that in the real world consciousness and learning are very close
companions. Thus the controversy about the Œnecessity‹ of
consciousness tends to interfere with a more subtle question
about the role consciousness plays in Œmost‹ cases of learning. We
will not review the learning controversy here; we raise it merely
to point out that whatever the answer may be to that question, it
does not negate the plain fact that most of the time when we want
to learn something we make ourselves conscious of the material to
be learned.
In order to avoid the unresolvable "zero-point" question, we
suggest a more answerable one: Do we need Œmore‹ conscious
involvement to learn Œmore‹ information? It seems likely that
routine and predictable information may be learned with minimal
conscious involvement. The more novelty we must absorb, the more
conscious experience we need. The evidence for this claim seems
to be widespread and non-controversial: the more words that need
to be memorized, the longer we must pay attention. The more
difficult and novel some material is, the more we time we must
spend being conscious of all its details and implications. And
so on. Figure 5.53 presents a theoretical curve describing this
situation. It shows an upward monotonic function between the
amount of information that is to be learned, and the amount of
conscious involvement needed to learn it. Notice that the zero
point of the curve is undefined, reflecting the difficulty of
deciding whether consciousness is a Œsine qua non‹ of learning. The<j <
figure suggests that we do not need to solve this problem in
order to make interesting claims about the relationship between
learning and consciousness.
----------------------------------------
Insert Figure 5.53 about here.
---------------------------------------
5.6 Some experimental predictions.
The key to theoretical success, of course, is making novel
predictions that work. Here are some possibilities.
5.61 The "cold dog" experiment.
There is evidence that unconscious (or at least,
unreportable) words may prime subsequent conscious processes.
Marcel (1983 ab) and others have shown that backward-masked
printed words, which are not reportable, still improve lexical
decision time for related words (i.e., the time that is needed
to decide whether some string of letters is a word or not). While
these results are apparently reliable, they have given rise to
great controversy (Cheesman & Merikle, 1984; Holender, 1986). The
debate is mainly about whether the unreportable words are truly
unconscious or not. That is, it is about the "zero point" of
consciousness, precisely the issue that is most difficult to
decide. We do not take a position on this issue, of course, but
GW theory does suggest an experimental prediction.
Briefly, when we have a very short unreportable exposure of
a word compound like "hot dog," it should prime lexical decisions
about related words, like "sausage". However, if the word
compound is novel, like "cold dog," it should not prime a related
term like "Huskie" or "frozen Fido". New word compounds require
conscious involvement to bring many different specialists to bear
on the problem of creating a coherent interpretation. There is
indeed one report in the literature supporting this hypothesis
(Greenwald, 198x).
<j <† 5.62 Other testable questions.
We have made the claim above that semantic satiation is one
source of evidence for the generality of redundancy phenomena. In
fact, it is difficult to do clean experiments with semantic
satiation (Amster, 1962; Esposito & Pelton, 1971). For example,
we do not know for sure that semantic satiation is an abstract
rather than a perceptual event, because when we repeat a word
over and over again, both the meaning and the perceptual stimulus
are repeated. To prove that semantic satiation is indeed semantic
we would have to repeat different synonyms or paraphrases with
the same conceptual meaning but different perceptual form, and
show that satiation occurs. Reddy and Newell (1974) cite more
than 100 paraphrases of a single sentence about a chess position.
If we repeat all l00 paraphrases, we should expect to find
semantic satiation if this is a truly conceptual, as opposed to
perceptual, phenomenon.
There appears to be almost no work on the issue of blindness
to conceptual presuppositions, although everyone must encounter
this phenomenon in everyday life. Measures of retrievability
could be easily used to test it (5.xx, 5.xx).
How is automatization of skills related to stimulus
habituation? They seem to be so similar, it is intriguing to
wonder about a connection. One possibility is that skilled
actions are guided by conscious and quasi-conscious goal images
(7.00). If that is so, perhaps automatization of skills simply
involves habituation of the relevant goal images. This hypothesis
may deserve further study.
Finally, it will be worth investigating the relationship
between information to be learned and the amount of conscious
involvement, using carefully designed stimuli with known
information content (e.g., Garner, 1974). This should cast light
on the heated issue of the relationship between consciousness and
learning without raising the inherent methodological difficulties
involved in seeking the "zero point" of consciousness.
5.7 Other implications.
5.71 Subliminal perception
There are two ways to model "subliminal" or unreportable
input in the GW model. First, if the input is routine, a
perceptual system may analyze it without recourse to the global<j <
workspace. The second possibility is that the GW may be used for
very rapid exchanges of information, and that linguistic and
recall systems that can report conscious experiences simply do
not have time to register this rapid global information. This is
similar to the Sperling (1960) phenomenon, where stimuli are
conscious briefly but cannot be recalled afterwards (1.xx). In
the second case a limited amount of novel processing could be
done, provided that other specialists in the system can react to
the global information more quickly than the linguistic and
recall specialists. Again, these alternatives are extraordinarily
difficult to test with our present methodology, but they are
worth pointing out.
We have now finished considering the meaning of a
"conscious experience." The following chapters will focus on the
relationships between multiple conscious events. This allows us
to deal with issues like problem incubation, voluntary control,
and conscious access to abstractions that are not experienced
qualitatively.
5.8 Summary.
This chapter has explored the fundamental phenomena of
habituation and automatization (Table 5.xx). We have argued that
all conscious contents must be informative, in that they trigger
widespread adaptive processes. Receiving specialists must feed
back their interest in the conscious content, so that they join
the coalition of systems that support the content. This is Model
3. It serves to put a conscious experience in a temporal context,
described as the "adaptation cycle." All conscious experiences,
it is argued, involve a stage of adaptation in which a defining
context has been accessed, so that the conscious information can
be understood; but not all degrees of freedom have been
determined. Consciousness occurs during the stage where the
remaining uncertainty in the defining context is being reduced.
After that point, adaptation has taken place, and repetition of
the same input will not result in a conscious experience. There
is thus an intimate connection between consciousness, adaptation,
information, reduction of uncertainty, redundancy, and context.
Information that fades from consciousness does not
disappear; rather, it serves to constrain later conscious
experiences. It may become part of a Œnew unconscious context,
within which later experiences are defined‹. One implication is<j <
that every event is experienced with respect to prior conscious
events: "Into the awareness of the thunder itself the awareness
of the previous silence creeps and continues..." as James says so
eloquently in our epigraph.
The more information we must adapt to, the longer we need to
be conscious of the material in order to create new contexts for
dealing with it; and the more new contexts we create, the more
our subsequent experiences will be reshaped. This suggests an
unconventional view of learning and development. Namely,
learning becomes a matter of developing contexts that cause us to
experience the same reality in new and different ways. We have
explored the evidence for this somewhat radical proposal.
In the upshot, this chapter suggests a third determining
condition for conscious experience, on a par with Œglobal
broadcasting‹ and Œinternal consistency‹ (Chapter 2). Namely: to be
conscious, a potential experience must be Œinformative‹. Even the
biological and personal significance of an event can be treated
as its informativeness in a goal context. This point allows GW
theory to deal with the issue of significance, a point that is
often neglected in the current cognitive literature.
Thus we are compelled to view even a single conscious
experience as part of a dynamic, developmental process of
learning and adaptation. Increasingly it seems that the system
underlying conscious experience is our primary organ of
adaptation.
In the following chapter we will explore the ways in which
contexts help to achieve goals. Much of our consciousness
involves thoughts about goals, ways of achieving goals, failures
to achieve them, and the like (Pope & Singer, 1978). In ordinary
life, as in the psychological laboratory, we are always asking
peole to do some task by given them a goal. "I would like you to
listen for a tone (goal), and to press this button (subgoal) with
your right hand (subgoal) as quickly as you can (subgoal) when
you hear one." But people are never conscious at any one time of
Œall‹ the details of motivation, levels of planning and motor
control, timing, testing of plans, and the like, which are needed
to reach even a simple goal. The bulk of our goalcrelated
processes are unconscious at any one time, even though they shape
our action and experience; that is to say, they are mostly
contextual. We show in the next chapter that some simple
assumptions about goal contexts and conscious events leads to an
understanding of the stream of consciousness --- the "flights"
and "perches" of the mind from one apparently unrelated
experience to another.