conceptual spaces and consciousness: integrating cognitive and affective processes
TRANSCRIPT
![Page 1: CONCEPTUAL SPACES AND CONSCIOUSNESS: INTEGRATING COGNITIVE AND AFFECTIVE PROCESSES](https://reader035.vdocuments.us/reader035/viewer/2022073016/5750948b1a28abbf6bb9f086/html5/thumbnails/1.jpg)
CONCEPTUAL SPACES AND CONSCIOUSNESS:
INTEGRATING COGNITIVE AND
AFFECTIVE PROCESSES
ALFREDO PEREIRA JÚNIOR
UNESP, State University of S~ao Paulo,
Institute of Biosciences, 18618-000,
Botucatu, SP, Brazil
LEONARDO FERREIRA ALMADA
UFG, Federal University of Goias/UNESP,State University of S~ao Paulo, 18618-000,
Botucatu, SP, Brazil
In the book \Conceptual Spaces: the Geometry of Thought" [2000] Peter Gärdenfors proposes anew framework for cognitive science. Complementary to symbolic and subsymbolic [connec-
tionist] descriptions, conceptual spaces are semantic structures ��� constructed from empirical
data ��� representing the universe of mental states. We argue that Gärdenfors' modeling can beused in consciousness research to describe the phenomenal conscious world, its elements and
their intrinsic relations. The conceptual space approach a®ords the construction of a universal
state space of human consciousness, where all possible kinds of human conscious states could be
mapped. Starting from this approach, we discuss the inclusion of feelings and emotions inconceptual spaces, and their relation to perceptual and cognitive states. Current debate on
integration of a®ect/emotion and perception/cognition allows three possible descriptive
alternatives: emotion resulting from basic cognition; cognition resulting from basic emotion, andboth as relatively independent functions integrated by brain mechanisms. Finding a solution for
this issue is an important step in any attempt of successful modeling of natural or arti¯cial
consciousness. After making a brief review of proposals in this area, we summarize the essentials
of a new model of consciousness based on neuro-astroglial interactions.
Keywords: Consciousness; conceptual spaces; correlations; methodology; homeomorphism.
1. Introduction
The scienti¯c study of subjectivity displays important lines of development in the last
50 years. In the human sciences, the qualitative methods of empirical research, using
a diversity of tools to collect and analyze data and developing its own validation
procedures, was progressively accepted by the scienti¯c community. Scienti¯c
methods derived from philosophy, as the phenomenological and dialectical ones, have
International Journal of Machine ConsciousnessVol. 3, No. 1 (2011) 127�143
#.c World Scienti¯c Publishing Company
DOI: 10.1142/S1793843011000649
127
Int.
J. M
ach.
Con
scio
us. 2
011.
03:1
27-1
43. D
ownl
oade
d fr
om w
ww
.wor
ldsc
ient
ific
.com
by 9
3.18
0.53
.211
on
08/1
4/13
. For
per
sona
l use
onl
y.
![Page 2: CONCEPTUAL SPACES AND CONSCIOUSNESS: INTEGRATING COGNITIVE AND AFFECTIVE PROCESSES](https://reader035.vdocuments.us/reader035/viewer/2022073016/5750948b1a28abbf6bb9f086/html5/thumbnails/2.jpg)
been incorporated into the methodology of these sciences. At the same time, several
e®orts have been made to correlate subjectivity with objective processes studied by
neurobiology and biophysics. Neural bases of psychological processes as learning,
memory formation, attention and emotion have been identi¯ed for several biological
species. The search for correlates of conscious experience has brought important
results, as the oscillatory synchrony in the thalamo-cortical system correlating with
conscious perceptual processes [see, e.g., Rodriguez et al., 1999]. However, while there
is remarkable progress in the identi¯cation of specialized brain regions involved in
several kinds of conscious processing, few advances have occurred in the under-
standing of how the informational content of brain states emerge as conscious lived
experiences.
A neurobiological explanation of the processing of conscious content would begin
by identifying what is there to be explained, or the explanandum, in the terminology
used by Hempel and Oppenheim [1948], Hempel [1965] and Popper [1963]. Such an
explanandum refers to all possible kinds of human conscious subjective experiences,
which can be mapped as the state space of human consciousness (for the use of the
notion of \state space" in this context, see Stanley [1999] and Fell [2004]). This task
depends on the analysis and categorization of empirical data, composed of reports of
individual human conscious experiences, implicitly assuming that subjectivity has a
universal structure.
In this paper, we acknowledge Peter Gärdenfors' proposal of constructing con-
ceptual spaces as a descriptive tool for a science of consciousness, corresponding to
the ¯rst step of de¯ning the explanandum of this science. We also pose the question of
how to integrate perceptual/cognitive with a®ective/emotional states in a conceptual
space that aims to describe the actual structure of human (and possibly human-based
arti¯cial) consciousness. We claim that a®ect/emotion and perception/cognition are
processed as relatively independent functions and integrated by a supplementary
mechanism.
A second step in the scienti¯c methodology for the study of human consciousness
is to de¯ne the explanans, the kind of biophysical process in the brain (and its
interactions with the body and environment) able to account for the processing of
conscious content. A coarse-grained mapping of brain regions involved in each kind of
conscious processing is not su±cient for this task; it is also necessary to identify how
information is encoded and processed by neuronal and glial cellular populations.
According to Gärdenfors [2000], one central property of conceptual spaces is the
existence of correlations between dimensions, a feature that would correspond to the
existence of large coherent patterns of activity in the brain. A positive matching of
such structures (the state space that describes the contents of human consciousness,
and the state space of biophysical information processing in the brain), displaying
isomorphic or homeomorphic inter-relations, would validate this methodology. In
this regard, we make a review of the literature on cognitive and emotional functions,
discussing how their integration can be described in conceptual spaces without
128 A. Pereira Jr. & L. F. Almada
Int.
J. M
ach.
Con
scio
us. 2
011.
03:1
27-1
43. D
ownl
oade
d fr
om w
ww
.wor
ldsc
ient
ific
.com
by 9
3.18
0.53
.211
on
08/1
4/13
. For
per
sona
l use
onl
y.
![Page 3: CONCEPTUAL SPACES AND CONSCIOUSNESS: INTEGRATING COGNITIVE AND AFFECTIVE PROCESSES](https://reader035.vdocuments.us/reader035/viewer/2022073016/5750948b1a28abbf6bb9f086/html5/thumbnails/3.jpg)
disrupting the intended correspondence with brain physiology. The results are
applied for the understanding of conditions to be satis¯ed by conscious arti¯cial
systems.
2. From Intersubjective Validation to Universal Conceptual Spaces
Frith et al. [1999] proposed a method to study the neural correlates of consciousness
that is based on the temporal co-occurrence of registered/measured brain activity
(using EEG, MEG, fMRI, PET-scanning, etc.) and conscious states reported by the
subject who is having his/her brain activity monitored. In our analysis of this
method, we make a distinction between conscious episodes instantiated by brain
activity and lived experiences involving brain-body-environment dynamical inter-
actions [Pereira Jr. and Ricke, 2009]. For instance, currently available \mind-reading"
technology [e.g., Kay et al., 2008] can access coarse-grained brain functional con-
¯gurations but cannot measure lived experiences. Partial accessibility of the latter to
the experimenter depends on verbal and/or non-verbal reports made by the human
subject. Therefore, what this methodology actually gets is a correlation between
measured brain activity and reports made by the subject about the content that he/
she experienced while the activity occurred.
The above procedures are too weak to determine a causal or law-like relationship
between brain states/processes and conscious states/processes, but su±ciently strong
to be considered as scienti¯c. A part of the weakness derives from the limitation of
isolated studies, where the correlations may be biased by cultural or individual
peculiarities. Therefore a possibility of adding a further strength to this methodology
is by crossing the results obtained by several subjects and several experimenters, in
order to identify the common aspects of reported human experiences.
The crossing of several experiments in the cognitive neuroscience of consciousness
makes possible an inter-subjective validation of the method. The rationale is that it
would be extremely unlikely that several human subjects, sharing the same class of
biological (genetic and phenotypic) constraints, interacting with the same kind of
experimental setting, and not being a®ected by any well-known psychopathology or
under e®ect of psychotropic drugs, would have di®erent lived experiences.
It is important to consider that one stance of inter-subjective validation always
occurs in any cognitive neuroscienti¯c experiment, when the experimenter makes a
confrontation of the subject's report and his/her own perception of a presented
stimulus (e.g., when the experimenter presents an object that he/she perceives to be a
blue circle, he/she expects the subject to report seeing a blue circle and not, for
instance, a red triangle). The idea of crossing subjective experiences is implicit in
Dennett's proposal of a hetero-phenomenological approach [Dennett, 1991], but his
approach was conceived as a refutation of ¯rst-person approaches to consciousness,
not as a procedure of inter-subjective validation as we are proposing here. The
original hetero-phenomenological method also lacks an appropriate framework to
plot the result of studies made by di®erent researchers analyzing di®erent subjects.
Conceptual Spaces and Consciousness 129
Int.
J. M
ach.
Con
scio
us. 2
011.
03:1
27-1
43. D
ownl
oade
d fr
om w
ww
.wor
ldsc
ient
ific
.com
by 9
3.18
0.53
.211
on
08/1
4/13
. For
per
sona
l use
onl
y.
![Page 4: CONCEPTUAL SPACES AND CONSCIOUSNESS: INTEGRATING COGNITIVE AND AFFECTIVE PROCESSES](https://reader035.vdocuments.us/reader035/viewer/2022073016/5750948b1a28abbf6bb9f086/html5/thumbnails/4.jpg)
This kind of framework is provided by Gärdenfors. It is composed of geometrical
tools (quality domains, similarity judgments as measure of proximity) used for the
construction of conceptual spaces. Therefore, an important advancement provided
by this new approach is to allow the progressive build-up of results of individual
hetero-phenomenological studies into what might be called the universal state space
of human consciousness.
The lived experience of an individual human subject, as long as it can be com-
municated to others, would correspond to a trajectory in this state space. Considering
that ¯rst-person lived experiences are partially available to empirical (third-person)
science by means of verbal or non-verbal reports, this framework can support the
interpretation of \qualitative" data (i.e., reports of lived experiences) in the context of
cognitive sciences. The state space model can be useful as a referential framework also
for introspective and phenomenological approaches.
More importantly, for this scienti¯c methodology the distinction of ¯rst- and
third-person perspectives is secondary, while keeping centrality from a philoso-
phical perspective. Perceptual concepts and \qualia" refer to the same entities,
suggesting the possibility of partial inter-translation of third-person and ¯rst-
person descriptions. If a set of common results is reported by di®erent subjects,
e.g., if they agree on the perceptual content elicited by a stimulus, the di®erence
between the perspectives disappears for practical purposes. As the term \qualia"
has a long and disputed history in philosophy, when possible we prefer to avoid its
usage and refer to the corresponding concepts. One of the problems is that a
\quale" includes both the subjective feeling (e.g., \what it is like for Adam to be
looking at a red apple") as well as apparently objective informational properties
(e.g., \the redness of red").
The partial conclusion we draw here is that the conceptual space approach is a
valuable tool for the science of human consciousness and practical applications, since
it a®ords the construction of a universal model representing the properties of the
\explanandum" of this science.
3. Conceptual Spaces and A®ective/Emotional States
Gärdenfors' approach is supported by empirical evidence from psychophysics, cog-
nitive psychology, linguistics and neuroscience. The use of data from cognitive
experimentation for the construction of perceptual spaces is possible because he
implicitly assumes a universal structure of human cognition. Then, the subjects'
reports of lived experiences in several experiments can be constructively mapped into
the theoretician's model.
A problem that appears for an approach to conscious processing based on con-
ceptual spaces is how to integrate a®ects and emotions with cognitive states. Since
a®ects/emotions are fundamental to conscious processes, they should be accounted
for by the conceptual space approach for this method being successful as a description
of the \explanandum" of consciousness science.
130 A. Pereira Jr. & L. F. Almada
Int.
J. M
ach.
Con
scio
us. 2
011.
03:1
27-1
43. D
ownl
oade
d fr
om w
ww
.wor
ldsc
ient
ific
.com
by 9
3.18
0.53
.211
on
08/1
4/13
. For
per
sona
l use
onl
y.
![Page 5: CONCEPTUAL SPACES AND CONSCIOUSNESS: INTEGRATING COGNITIVE AND AFFECTIVE PROCESSES](https://reader035.vdocuments.us/reader035/viewer/2022073016/5750948b1a28abbf6bb9f086/html5/thumbnails/5.jpg)
First, it is necessary to clarify what we mean by \a®ect" and \emotions", and how
they di®er from cognitive states. \A®ect" refers to a class of subjective feelings, such
as those related to basic physiological functions (hunger, thirst, heat, pain, body
pleasure, etc.) and psychological states (love and hate, happiness, sadness, etc.).
\Emotion" refers to feelings related to the objects of action in a bio-psycho-social
context, e.g., anger (of something), anxiety (for something), fear (of something), etc.
The distinction is surely a fuzzy one.
Conceptual spaces are de¯ned as \theoretical entities that can be used to explain
and predict various empirical phenomena concerning concept formation" [Gärden-fors, 2000, p. 31]. They are made of quality dimensions, representing the qualities of
objects, like weight, brightness and pitch. The proximity or distance between
dimensions is established by judgments of similarity, which can be made in the
context of scienti¯c experimentation. Such judgments establish intra- and inter-
dimensional correlations. How could a®ects and emotions be included in conceptual
spaces? To answer this question, three issues are brought into consideration:
(i) A®ects and emotions are lived experiences, not concepts. One problem with the
proposal of a mathematical \qualia space" by Stanley [1999] was the con°ation
of concepts and experiences. Universal types of conscious experience can be
represented by concepts (which are indicated by terms as \pain" and \pleasure",
in the above examples). Conceptual spaces, of course, are not made by lived
experiences themselves, but by their concepts. In order to achieve a realistic view
of the human mind or any arti¯cial system that simulates or reproduces
activities of the human mind, the relations between concepts should have
a homeomorphism with actual relations. This correspondence is assured, in
Gärdenfors' case, by the method of construction of the model;
(ii) Conceptual spaces, as they were presented by Gärdenfors [2000], are heavily
based on perceptual and cognitive processes. The symbol grounding problem is
solved since perceptual domains \are tied to sensory input channels and hence
what is represented in these domains has some correspondence with the external
world" [Gärdenfors, 2000, p. 43, 44]. The learning problem of connectionist
networks is also approachable, since it provides an understanding of concept
formation by means of judgments of similarity. The frame problem is solved
when information is not conceived as modality-free, but sorted into domains.
The properties of objects are conceived as perceptual invariances located in
conceptual spaces. While properties are based on one quality domain, perceptual
concepts are based on several domains. Therefore, they aremultimodal instead of
amodal; e.g., the concept of \apple" involves the color, shape, texture, taste,
fruit and nutrition domains, see Gärdenfors [2000, p. 103]. How to apply the
same strategy of argumentation — based on the embodiment of processes — to
a®ective and emotional phenomena? In order to answer this question, it is
necessary to examine how scientists conceive the process by which the brain
supports lived experiences and discuss the inclusion (or not) of other parts of the
Conceptual Spaces and Consciousness 131
Int.
J. M
ach.
Con
scio
us. 2
011.
03:1
27-1
43. D
ownl
oade
d fr
om w
ww
.wor
ldsc
ient
ific
.com
by 9
3.18
0.53
.211
on
08/1
4/13
. For
per
sona
l use
onl
y.
![Page 6: CONCEPTUAL SPACES AND CONSCIOUSNESS: INTEGRATING COGNITIVE AND AFFECTIVE PROCESSES](https://reader035.vdocuments.us/reader035/viewer/2022073016/5750948b1a28abbf6bb9f086/html5/thumbnails/6.jpg)
body and parts of the environment in the process that generates a®ects/
emotions;
(iii) The location of a linguistic term in the conceptual space may vary, depending on
the context of the utterance. Conceptual spaces can also be used as a basis for
cognitive semantics, replacing realist approaches (both extensional and inten-
sional). Gärdenfors proposes that \meanings of linguistic expressions are mental
entities" [Gärdenfors, 2000, p. 154], at the same time holding that conceptual
structures of di®erent individuals become attuned to each other: \via successful
and less successful interactions with the world, the conceptual structure of an
individual will adapt to the structure of reality" [Gärdenfors, 2000, p. 156].He assumes a \socio-cognitive" position, accounting for social power in°uences
on conceptual structures. These constraints surely apply to a®ect/emotion,
mostly to the latter, since it is by de¯nition related to events in the body and
environment.
The construction of a universal conceptual space representing the universe of
conscious contents is proposed to be the ¯rst step in the methodology of a science of
consciousness. It requires collection of verbal and non-verbal reports of ¯rst-person
experiences correlated to measured brain and body activity; inter-subjective vali-
dation of the reports, and construction of a conceptual space of conscious contents.
An analysis of the structure of this space may further guide the search for arti¯cial
reproductions of consciousness.
The ¯rst step should be accompanied by a second one, the theoretical modeling of
brain activity at multiple spatio-temporal scales, composing a state space of brain
functions. This strategy implies that homeomorphisms [Fell, 2004] between the
structure of conscious contents and the structure of brain/body activity are going to
be found by means of a matching of the state space of human consciousness with the
corresponding state space of human brain functions.
Finding a solution for the issue of integration of emotional and cognitive pro-
cesses is an important step in any attempt of successful modeling of natural or
arti¯cial consciousness. A ¯rst move in the direction of including a®ective/
emotional processing as a fundamental component of consciousness is the Somatic
Marker Hypothesis (SMH), by Dam�asio and colleagues. According to the defenders
of the hypothesis, a®ects/emotions are related to somatic states, being preceded
and followed by cognitive processes. The latter include both basic and higher brain
processes, from simple perceptions and thoughts to attention, evaluation, and
decision-making. SMH considers a®ect/emotion as \a collection of changes in a
body state connected to particular mental images (thoughts) that have activated a
speci¯c brain system" [Dam�asio, 1994, p. 159]. Feelings of emotions are ¯rst and
foremost considered to be about the body: \they o®er us the cognition of our
visceral and musculoskeletal state as it becomes a®ected by pre-organized mech-
anisms and by the cognitive structures that have developed under their in°uence"
[Dam�asio, 1994, p. 159].
132 A. Pereira Jr. & L. F. Almada
Int.
J. M
ach.
Con
scio
us. 2
011.
03:1
27-1
43. D
ownl
oade
d fr
om w
ww
.wor
ldsc
ient
ific
.com
by 9
3.18
0.53
.211
on
08/1
4/13
. For
per
sona
l use
onl
y.
![Page 7: CONCEPTUAL SPACES AND CONSCIOUSNESS: INTEGRATING COGNITIVE AND AFFECTIVE PROCESSES](https://reader035.vdocuments.us/reader035/viewer/2022073016/5750948b1a28abbf6bb9f086/html5/thumbnails/7.jpg)
However, SMH defenders also hold that emotional appraisal can interfere with
and even support higher cognitive functions, as decision-making: \individuals make
judgment not only by assessing the severity of outcomes and their probability of
occurrence, but also and primarily in terms of their emotional quality" [Bechara
et al., 2000, p. 305]. This part of their theory leads to what would be considered as
bad explanatory circularity, causing possible inconsistencies criticized by Bennett
and Hacker [2005] and Rolls [2000].
Recognizing the embodiment of emotion would require more than merely mapping
brain activity, since emotion is related to the actions of the living individual's body in
the environment. While the content of each feeling can be considered to be the
product of a specialized, epigenetically determined brain circuit or module, the
content of emotions vary according to the context (state of the body and environ-
ment). In this regard, SMH brings an important contribution to the understanding of
embodiment of emotions when taking into account the reciprocal signaling of brain
and body. In spite of a possible explanatory circularity in its framework, the SMH
may contribute to the description of integrative functions in the universal conceptual
space of consciousness, since lived experiences should necessarily include the dyna-
mical relation of brain, body and environment.
A second position, illustrated by the work of LeDoux [1996] and Panksepp [1998],
assumes the independence ��� and, in many cases, a predominance ��� of a®ects/
emotions on cognitive activity. This position is able to avoid most of the criticisms
made to SMH by Rolls, since it does not attempt to derive the content of feelings and
emotions from somatic activity, but take them as brain primitives. However, the
predominance of emotion over cognition con°icts with empirical evidence gathered
by SMH defenders, and contradicts theoretical argumentation by Bennett and
Hacker. The latter authors are opposed to Dam�asio and colleagues only regarding the
dependence of decision-making on somatic markers and related emotions.
Considering available evidence reviewed by the above authors, we suggest that it
could be encompassed by a third possibility of mapping the dynamics of cognitive and
emotional processing. Perceptual/cognitive and a®ective/emotional functions are
primitive brain functions, being able to occur both implicitly (unconsciously) and
explicitly (consciously). Modeling of perceptual/cognitive systems in conceptual
spaces was largely advanced by Gärdenfors [2000], while modeling of emotional
systems was advanced by Panksepp [1988], in a pioneering work that could equally be
translated to conceptual spaces.
According to Panksepp [1998], our cognitive representation of the world merges
within the processing of a®ective states. A®ective states are related to the awakening
of primitive emotional command circuits in the whole extended neural network,
including the operational systems he calls (using capitol letters) \SEEKING, FEAR,
RAGE, LUST, CARE, PANIC, and PLAY" [Panksepp, 1998; 2005]. These would
be fundamental dimensions of the human a®ective conceptual space that interact
with perceptual/cognitive dimensions. Cognitions re-evoke feelings in function of our
Conceptual Spaces and Consciousness 133
Int.
J. M
ach.
Con
scio
us. 2
011.
03:1
27-1
43. D
ownl
oade
d fr
om w
ww
.wor
ldsc
ient
ific
.com
by 9
3.18
0.53
.211
on
08/1
4/13
. For
per
sona
l use
onl
y.
![Page 8: CONCEPTUAL SPACES AND CONSCIOUSNESS: INTEGRATING COGNITIVE AND AFFECTIVE PROCESSES](https://reader035.vdocuments.us/reader035/viewer/2022073016/5750948b1a28abbf6bb9f086/html5/thumbnails/8.jpg)
past experiences and history of conditioning, thus interacting with emotional
self-regulation.
The dynamics of emotional systems tends to be more egocentric and uncondi-
tionally a®ective than cognitive functioning, since the goal of cognitive processes is
generally to provide more subtle solutions to problems posed by emotional arousal. In
hominids, the evolution of certain higher symbolic abilities provided ways for
organisms to solve con°icts that are very di±cult from an exclusively emotional
perspective. The increasing in°uence of cortical functions provides cognitive
resources to solve con°icts during human maturation. This evolution was critically
\guided by the pre-existing neurobiological exigencies of organisms" [Panksepp,
2001, p. 144]. In fact, these exigencies are related to subcortical emotional and
motivational abilities \which are generally more similar among living mammalian
species than their higher cortico-cognitive functions which have diverged more con-
siderably" [Panksepp, 2001, p. 144].
A®ective states re°ect evolutionary value codes. In this sense, the projection of
feelings onto environmental events and objects was one of the simplest ways for
evolution to persistently guide perceptual priorities of the cognitive apparatus. This
type of interaction occurs by means of global neurodynamic processes in the brain,
making possible that cognitive problems could be ameliorated simply by adjusting
the underlying emotional feelings, and vice versa. In other words, a®ective states of
consciousness \may, quite simply, be among the most robust and e®ective ways to re-
channel cognitive resources" [Panksepp, 2001, p. 152]. Cognitive forms of con-
sciousness were evolutionarily grounded \on the prior evolution of a®ective forms of
consciousness, which inform organisms what it might be worth thinking about"
[Panksepp, 2001, p. 152].
When brain functions occur consciously, there must necessarily be an integration
to de¯ne their access to the coordination of covert and overt behavior. This inte-
gration possibly happens in some kind of \Global Workspace" [Baars, 1997] embo-
died in brain activity. One possible implementation of the workspace is by means of
neuro-astroglial interactions, as recently suggested by one of us [Pereira Jr. and
Furlan, 2009; 2010]. In this approach, each brain circuit or functional module is
considered to be specialized for a speci¯c function, while the integration occurs by
means of the operation of a supplementary mechanism that involves virtually all
specialized circuits/modules, but is not identi¯ed with any of them.
4. From Arti¯cial Intelligence to Arti¯cial Consciousness
The project of \strong" Arti¯cial Intelligence was to construct machines able of
performing human cognitive functions. Critics soon noted that aspects of these
functions could not be performed by Turing machines; some of them also predicted
that these functions would not be reduced to digital computations, because of
several di±culties (as the combinatorial complexity and frame problems). The same
objections apply to projects of arti¯cial consciousness, which also poses two
134 A. Pereira Jr. & L. F. Almada
Int.
J. M
ach.
Con
scio
us. 2
011.
03:1
27-1
43. D
ownl
oade
d fr
om w
ww
.wor
ldsc
ient
ific
.com
by 9
3.18
0.53
.211
on
08/1
4/13
. For
per
sona
l use
onl
y.
![Page 9: CONCEPTUAL SPACES AND CONSCIOUSNESS: INTEGRATING COGNITIVE AND AFFECTIVE PROCESSES](https://reader035.vdocuments.us/reader035/viewer/2022073016/5750948b1a28abbf6bb9f086/html5/thumbnails/9.jpg)
additional problems:
(a) What is required for an agent to be conscious, besides information processing?
(b) What is the (arti¯cially reproducible) biophysical mechanism su±cient to
execute it?
An attempt to overcome initial di±culties is the \embodied and embedded"
paradigm of cognition, which claims that the strategy of implementing computations
based on internal databases should be substituted by an analogue of brain-body-
environment interactions. One of the main attempts to bring this idea to reality was
Rodney Brooks' robot \Cog", which has a central processing unit embedded in a
human-like half-body and was \educated" by a woman playing the role of \mother"
in a human-like environment [Brooks et al., 1998].
Why would a robot like Cog not, in principle, become a conscious agent? More
simply, why a thermostat that measures and controls room temperature does not feel
hot or cold? The answer, of course, depends on the concept of consciousness that is
assumed. Before making a speci¯c proposal (see next section), we would like to review
some alternatives presented by several authors.
A ¯rst line of reasoning is that not any representation is conscious, but only a
subclass ��� explicit representations. The problem here is to identify what features
make a representation explicit for the agent. For human beings, explicitness is often
related to symbolic and/or linguistic formulation of content. This line of reasoning
leads to di±cult issues of syntax and semantic processing, as those discussed by
Mandik [1999], leading to the suggestion of a \procedural psychosemantics" that
evaluates properties of introspective content of experiences (\qualia") regarding their
e®ects (instead of their causes). This reasoning does not apply to all qualia, but to
spatial properties related to motor control. In this regard, the theory of a \retinoid
system" by Trehub [1991] describes a brain subsystem able to construct an egocentric
three-dimensional transparent representation of the world. Like inMandik's approach,
this proposal is limited to the spatial structure of perceptual consciousness. The
retinoid system is a model of brain mechanisms supporting our egocentric spatial
representation of the world, leaving open the explanation of other dimensions of the
conceptual space and corresponding brain mechanisms.
Another possible requirement is the capacity of elaborating self-referential
representations, as in High-Order Thought theories of consciousness [Rosenthal,
2002; 2004]. Rosenthal [2002] begins by distinguishing between \creature con-
sciousness" and \state consciousness", assuming that animals like frogs and turtles
are creature-conscious while being fully state-unconscious (i.e., all their mental states
may well not be conscious). Arguing from what seems to ¯gure in the phenomenon of
state consciousness, he argues that \being aware of a mental state…is not a su±cient
condition for the state to be conscious". In his view, consciousness consists in a
thought to the e®ect that one is in the state in question, and the thought must not
rely on any inference. This holds for both conscious intentional states and conscious
Conceptual Spaces and Consciousness 135
Int.
J. M
ach.
Con
scio
us. 2
011.
03:1
27-1
43. D
ownl
oade
d fr
om w
ww
.wor
ldsc
ient
ific
.com
by 9
3.18
0.53
.211
on
08/1
4/13
. For
per
sona
l use
onl
y.
![Page 10: CONCEPTUAL SPACES AND CONSCIOUSNESS: INTEGRATING COGNITIVE AND AFFECTIVE PROCESSES](https://reader035.vdocuments.us/reader035/viewer/2022073016/5750948b1a28abbf6bb9f086/html5/thumbnails/10.jpg)
qualitative states, such as sensations. This is not an easy idea to implement in
arti¯cial systems, since before being able of self-reference it is necessary for the system
to have a Self ! This result is expected ��� in the long run ��� for evolutionary com-
puting paradigms, but there is no guarantee that it will occur. Other di±culties are
that the operation of self-reference is related to logical paradox, and possibly absent
in non-numan animals.
One scienti¯cally promising project is the attempt to quantify conscious inte-
grated information and related subjective experiences [Tononi, 2005; Balduzzi and
Tononi, 2009]. Tononi's theoretical framework is based on two ideas: that con-
sciousness is \informative" in the sense that any conscious state is a \reduction of
uncertainty" (it rules out many alternatives); and that conscious states integrate
brain information by means of causal interactions among its parts. He developed a
framework to measure the quantity of integrated information (phi) in brains that
may be useful for medical diagnosis and technological purposes, such as measuring it
in arti¯cial systems. A new development of the theory includes an approach to the
qualities of subjective experience (\qualia").
What would be the relation of integrated information with qualitative subjective
experience? Balduzzi and Tononi explain that \for integrated information to be high,
a system must be connected in such a way that information is generated by causal
interactions among rather than within its parts… The set of all submechanisms of the
system… conveniently captures all possible combinations of causal interactions"
[Balduzzi and Tononi, 2009, p. 5]. Then they introduce the concept of \informational
relationship" to represent \di®erences that make a di®erence" to the system. How
does a causal interaction \make a di®erence"? A condition of possibility is the
existence of a pre-existing conceptual space where entanglements (i.e., integrated
information relations) can be speci¯ed: \a fundamental property of q-arrows
(informational relationships ��� APJ/LA) is their entanglement… the extent to
which an informational relationship does not reduce to its component relationships
(sub-q-arrows)" [Balduzzi and Tononi, 2009, p. 6]. Each \quale" corresponds to an
activation of the conceptual space (\qualia space", in their terminology) by causal
brain mechanisms; therefore, contrary to the model presented by Stanley [1999],
subjective experiences would not pre-exist in qualia space, but are constructed from
brain activity. Although mathematically well built, both models still lack biophysical
grounding. We suggest that the proposal reviewed in the next section provides what
is missing to understand the brain basis of information integration and/or to
implement the concept in an arti¯cial agent.
A sophisticated interdisciplinary approach is the \physics of mind" proposed by
Perlovsky [2009]. Conscious processes are conceived as the matching of forms at
several levels, e.g., from retina to prefrontal neurons in the visual system. The initial
vague patterns that drive top-down signaling are innate (given by the Knowledge
instinct). At some point in human ontogenesis, this process includes language circuits
(also innate ��� the Language instinct), which allows us to use abstract forms that are
136 A. Pereira Jr. & L. F. Almada
Int.
J. M
ach.
Con
scio
us. 2
011.
03:1
27-1
43. D
ownl
oade
d fr
om w
ww
.wor
ldsc
ient
ific
.com
by 9
3.18
0.53
.211
on
08/1
4/13
. For
per
sona
l use
onl
y.
![Page 11: CONCEPTUAL SPACES AND CONSCIOUSNESS: INTEGRATING COGNITIVE AND AFFECTIVE PROCESSES](https://reader035.vdocuments.us/reader035/viewer/2022073016/5750948b1a28abbf6bb9f086/html5/thumbnails/11.jpg)
not innate (\memes"). The matching process applies to cognitive functions of other
animal species. Conscious representations are those that become crisp along the
matching processes.
Perlovsky's model is not strictly Kantian, since in the faculty of understanding
described in Critique of Pure Reason there is no matching of internal (a priori) and
external (empirical) forms. All forms are a priori ��� experience only gives us the
matter to ¯ll internal templates of space/time/causation/substance, etc. In Per-
lovsky's model, called \Dynamic Logic", the matching process a®ords an interaction
of externally originated and internal information patterns, avoiding the problem of
combinatorial complexity by means of an adequate use of Fuzzy Logics (instead of
the Principle of the Excluded Middle).
A criticism we make is about the relation of crispness and a®ective/emotional
states regarding consciousness. Possibly there is not a necessary connection between
the proposed con°ict-solving role of consciousness and crispness. A solution of a
con°ict of instincts/re°exes may be just choosing one of the vague unconscious
alternatives and abandoning the others. Of course, in some cases it may be better to
make a synthesis that uses parts of di®erent con°icting alternatives, but there is no
reason to assume that the synthesis has to be crisper than the original alternatives. In
fact, it may be even more vague if based on Fuzzy reasoning.
More importantly, we did not see a fundamental role for feelings in the conduction
of the process that leads to consciousness. Emotional feeling (as aesthetical feeling) is
regarded as a byproduct of good matching, not as a factor for selection of forms to
reach consciousness. It seems the model assumes ��� like most of Modern Philosophy
and contemporary Cognitive Science ��� that we are rational beings driven by cog-
nitive processes. We would better suggest that ��� besides the Language and the
Knowledge instincts assumed by Perlovsky ��� we also have A®ective instincts, as
argued in the context of Jaak Panksepp's A®ective Neuroscience (following Cannon's
and McLean's approach, also assumed by LeDoux and others).
If the a®ective instincts are introduced as fundamental ��� together with the
knowledge and language instincts ��� then the relation between crispness and con-
sciousness would have to be rede¯ned, since this property would not apply to feelings.
In fact, feelings are not \representations" measured as being more or less accurate
relatively to an intentional object or process. If feelings are considered to be
important players in the consciousness game, the crispness of representations
would not be the main parameter in the construction of scales to measure degrees of
consciousness.
In this section, we conclude for the need of an approach that goes beyond
SMH, inserting a®ective operational systems as fundamental dimensions of the
conceptual space. A question remains about the nature of a®ective states in the brain,
and how to implement it in arti¯cial systems. The next section is about a new
approach to the brain mechanisms underlying a®ective states and their putative
biophysical nature.
Conceptual Spaces and Consciousness 137
Int.
J. M
ach.
Con
scio
us. 2
011.
03:1
27-1
43. D
ownl
oade
d fr
om w
ww
.wor
ldsc
ient
ific
.com
by 9
3.18
0.53
.211
on
08/1
4/13
. For
per
sona
l use
onl
y.
![Page 12: CONCEPTUAL SPACES AND CONSCIOUSNESS: INTEGRATING COGNITIVE AND AFFECTIVE PROCESSES](https://reader035.vdocuments.us/reader035/viewer/2022073016/5750948b1a28abbf6bb9f086/html5/thumbnails/12.jpg)
5. A Viable Implementation of Universal Conceptual Spacesin Neuro-Astroglial Networks
In a recent work, Pereira Jr. and Furlan [2010] propose a new model of consciousness
that includes a central role for astrocytes. The principles of this model can be
extended to arti¯cial systems. The model departs from the sketch of a large scale Ion
Trap Quantum Computer (ITQC) adapted to biological conditions [Pereira Jr.,
2007]. In quantum computing models, it is assumed that the trajectory of particles,
under well known initial and boundary conditions (quasi-isolated micro-system fro-
zen to the ground state) is determined by fundamental physical laws. As these laws
are temporally invariant, computations carried in this kind of system are considered
to be reversible. In a biological macro-system, there are at least three factors that
indicate an incompleteness of quantum theory to describe the dynamics:
(a) The process of decoherence (or objective reduction or \collapse") of the wave-
function, attributed (by several authors) to supplementary factors;
(b) The many-body problem (discussed by Poincar�e as \three-body problem"),
leading to an impossibility of deterministic calculation of the trajectory. At this
point, Statistical Mechanics take the place of Quantum Theory in the expla-
nation of observed macro-phenomena;
(c) The Second Law of Thermodynamics, causing the evolution of the macro-system
to be irreversible.
Initial conditions of living systems include a database, the DNA, from which
Biological Maxwell Demons (enzymes) are built. The Second Law requires that the
work of enzymes, reducing internal entropy, must be compensated by means of
consumption of useful energy from the environment, implying that these systems
must consume food to survive. Both DNA information present in the initial state and
the necessity of consumption of low entropy from the environment are not theorems
of Quantum Theory. These conditions, necessary for the existence of life, are also
related to the formation of coherent calcium waves in astrocytes, since a class of
proteins crucial for brain activity, the ligand-gated ion channels (e.g., NMDA
channels) operate like Biological Maxwell Demons, controlling the °uxes of ions
through cellular membranes and compartments.
We conceive large calcium waves as composed of an ensemble of small standing
waves, each one inside an astrocyte microdomain. Their connection, forming a larger
pattern of coherent activity, is made by ATP signaling (in the case of the \domino
e®ect", see Pereira Jr. and Furlan [2010]) and by neuronal synchronous ¯ring (in the
case of the \carousel e®ect", see Pereira Jr. and Furlan [2010]). Microdomains are
subcellular, but the domino and carousel e®ects are assumed to transpose their
patterns to a collective of cells, thus composing a fractal multiscale pattern.
Also quantum entanglement of calcium ions possibly plays a role in intercellular
communication, but this communication can occur only after the waves are formed.
With entanglement, when a small wave inside a microdomain is formed, another
138 A. Pereira Jr. & L. F. Almada
Int.
J. M
ach.
Con
scio
us. 2
011.
03:1
27-1
43. D
ownl
oade
d fr
om w
ww
.wor
ldsc
ient
ific
.com
by 9
3.18
0.53
.211
on
08/1
4/13
. For
per
sona
l use
onl
y.
![Page 13: CONCEPTUAL SPACES AND CONSCIOUSNESS: INTEGRATING COGNITIVE AND AFFECTIVE PROCESSES](https://reader035.vdocuments.us/reader035/viewer/2022073016/5750948b1a28abbf6bb9f086/html5/thumbnails/13.jpg)
small wave in a spatially distant region of the astrocyte network can be a®ected,
without any particle from the initial wave being actually transported to the second.
However, quantum entanglement cannot explain the formation of waves, since this
process requires mechanical causation. The two, domino and carousel e®ects, are
intended to explain the formation of waves; after formed, quantum entanglement
may occur within them. Considering both e®ects, the time needed to produce a large
coherent calcium wave is close to the time needed to form small calcium waves in
microdomains.
Our claim is that the brain substrate of feelings cannot be a FM (Frequency
Modulated), but has to be an AM (Amplitude Modulated) pattern. FM is very good
to transmit information and reduce noise. The problem is that it requires a decoder to
read the message. In the brain, FM a®ords a digital-like encoding of sensory and
motor signals. In the ¯rst case, the receiver is a CNS (Central Nervous System)
neuron; in the second case, a muscle or gland. However, the idea of FM information
transmission to consciousness would require an homunculus, which was resuscitated
by Crick and Koch [2003].
We conceptualize our feelings (e.g., pains and pleasures) as homeomorphic to large
and coherent AM astrocytic calcium waves, instead of Morse-like sequences of pulses
in millions of neurons. As the astroglial calcium wave is conceived as the ¯nal stage of
conscious processing, there is no need of a receiver. The ¯nal step would be an analog
reproduction (a \presentation", not a \representation") of the properties of the
world, by means of an AM waveform, which feedsback on the whole living individual.
Therefore, at the end of the line there should be a collective large-scale AM wave. The
only place for it in the brain is the astroglial network.
Neurons process information and trigger muscle/gland action without having the
feeling of what is going on. The dendritic graded potential is ¯ltered by the axon
hillock to generate a Frequency Modulated (FM) signal along the axon. Synapses
receives only this signal; they do not have access to the dendritic AM pattern. The
di®erence for the astrocyte network is the possibility of a collective AM wave. Phi-
bram [1991] and Hamero® [2010], appealed to dendritic ¯elds as substrates of con-
sciousness, but how would one dendritic tree transmit its AM pattern to others? The
solution found by Hamero® is to make electric synapses responsible for this task, but
the attempt fails because these synapses only transmit electric current that °ows to a
segment of a dendrite, not the information pattern present in the graded potential as
a whole.
All brain areas contain both neurons and astrocytes. Their cognitive functions are
mediated by glutamatergic transmission. Neurons operate as specialized information
processing detectors/¯lters/relays, but do not convey a feeling about the content of
the information they process. For example, in the amygdala, neurons ¯re to a®ective/
emotional salient/relevant stimuli, but the corresponding feelings are not instan-
tiated by them [see Pessoa and Adolphs, 2010]. Pereira Jr. and Furlan [2010] propose
that feelings are instantiated by a large astroglial network connected to the
Conceptual Spaces and Consciousness 139
Int.
J. M
ach.
Con
scio
us. 2
011.
03:1
27-1
43. D
ownl
oade
d fr
om w
ww
.wor
ldsc
ient
ific
.com
by 9
3.18
0.53
.211
on
08/1
4/13
. For
per
sona
l use
onl
y.
![Page 14: CONCEPTUAL SPACES AND CONSCIOUSNESS: INTEGRATING COGNITIVE AND AFFECTIVE PROCESSES](https://reader035.vdocuments.us/reader035/viewer/2022073016/5750948b1a28abbf6bb9f086/html5/thumbnails/14.jpg)
amygdala's astrocytes. In this view, it is not the amygdala that feels fear, or the insula
that feels pain or pleasure; the conscious subject is the whole brain/body of the living
individual, having the astroglial network as the main connector (Master Hub)
between his/her subsystems (blood, cerebral °uid, muscles, neurons).
According to model developed by Pereira Jr. and Furlan [2010], the Slow Cortical
Potential [He and Raichle, 2009] involved in default networks, as well as ERP waves
associated to conscious processing (P300, N400, etc.) would be mediated by the
astroglial network. Of course, scalp EEG does not tell us if the potential is mediated
by neuronal or astroglial transmission, but the relative slowness and concentrated
power of the wave suggests that it is mediated by the astroglial network instead of
neuron axon bundles (if so, the wave would be faster and have less impact on the EEG).
What would be the neurocentric interpretation of how the slow cortical potential
moves along brain tissue? The standard explanation is that the potential would move
along super¯cial layers of the cortex as a result of sequential excitation and inhibition
of columns. Thalamic glutamatergic input to a pyramidal neuron in deeper layers
elicit both a feedback to thalamus and an excitatory signal to super¯cial layers of the
same column. Excitation of this column leads to the excitation of inhibitory neurons
in adjacent columns, which inhibit the pyramidal neuron. Meanwhile, the super¯cial
layer neurons have propagated the excitation to another column, where the whole
cycle occurs again, and so on. The propagation would be by means of horizontal
cortico-cotical connections (endogenous) or by means of thalamic guidance (in the
case of an external stimulation). Magnetoencephalography would detect this move-
ment of the excitatory potential, while the EEG only detects the outcome.
The propagation of waves measured by scalp EEG by means of dendritic columns
is a hypothesis made to save the phenomena, but the hypothesis has never been really
proven. The clearest theoretical fact is that dendritic ¯elds in tripartite synapses
activate astrocytes and vice versa (astrocyte calcium waves feedback on neurons).
Therefore, the reasonable conclusion is that any slow cortical potential must involve
both dendritic ¯elds and astroglial calcium waves.
One most interesting test to check which one is more important to consciousness is
to identify which one is absent in non-REM sleep. An important result found by
Tononi and colleagues [Massimini et al., 2010] is compatible with the idea that lack of
connectivity during non-REM sleep results from deactivation of the astrocyte net-
work. Neurons continue to produce graded and action potentials during this phase of
sleep, the main di®erence being that they synchronize in slower frequencies (3Hz;
Delta). Pereira Jr. and Furlan [2009] hypothesized that the relevant e®ect of Delta for
unconsciousness is on astrocytes; with 3Hz, coherent large calcium waves are not
formed in the astrocyte network. However, a neurocentric explanation is also poss-
ible. Considering changes of concentration of available transmitters during slow-wave
sleep, it may be the case that one of the steps in the chain of neuronal events is
blocked. New results and computer simulations may helpfully indicate the causes of
reduced cortical connectivity during non-REM sleep.
140 A. Pereira Jr. & L. F. Almada
Int.
J. M
ach.
Con
scio
us. 2
011.
03:1
27-1
43. D
ownl
oade
d fr
om w
ww
.wor
ldsc
ient
ific
.com
by 9
3.18
0.53
.211
on
08/1
4/13
. For
per
sona
l use
onl
y.
![Page 15: CONCEPTUAL SPACES AND CONSCIOUSNESS: INTEGRATING COGNITIVE AND AFFECTIVE PROCESSES](https://reader035.vdocuments.us/reader035/viewer/2022073016/5750948b1a28abbf6bb9f086/html5/thumbnails/15.jpg)
6. Concluding Remarks
In our view of brain instantiation of the universal conceptual space of human con-
sciousness, conscious states are conceived as a conjoint product neuronal and astroglial
activity. Paraphrasing the title of Dam�asio's book, consciousness is \the feeling" (or
sentience) of \what happens" (information contents embodied in spatially distributed
neuronal activity). Sentience is supported by wavelike processing in astrocytes, while
awareness (in the sense of information processing) would depend on digital-like infor-
mation transmission by neurons. Neuronal dendritic ¯elds are wavelike, but axonal
transmissions are not. Dendritic ¯elds contain the information that becomes conscious,
but consciousness occurs only when these patterns are integrated by the astrocytic
network into a wavelike unity.
Consciousness is conceived as astroglial integration of information contents carried
by neurons. In this context, we make terminological equivalences: [astrocyte analog
wavelike processing� ¼ ½information integration� ¼ ½feeling� ¼ ½sentience�. Then
consciousness is ½ðastroglialÞ sentience of ðneuronalÞ awareness� ¼ [the feeling of the
information content] ¼ ½\the feeling of what happens"�. Neither neuronal awareness
of information nor astroglial sentience (e.g., unconscious emotions) are fully conscious.
When they get together, higher degrees of consciousness occur. Speci¯city is neuronal.
Di®erent features are represented by means of activation of neuronal receptive ¯elds.
Astrocytes do not reproduce neuronal information with ¯delity. The astrocytic
wavelike response conveys the feeling, which is intrinsically di®use.
In the above view, exclusive neuronal or astroglial activation would correspond to
lower degrees ��� proto-consciousness ��� while conjoint activation would correspond
to higher degrees ��� full consciousness, i.e., awareness with a feeling. \The feeling" is
also similar to Thomas Nagel's \what's like to be" that was important for Chalmers'
reasoning [Chalmers, 1995; 1996].
According to these guidelines, the generation of arti¯cial conscious agents
would require both computational and a®ective mechanisms to reproduce neuronal
functions ��� specialized digital processing of features of information ��� and astro-
glial functions ��� integration of such a distributed processing and generation of a
feeling about the message carried by the signals. These mechanisms are physically
di®erent; the ¯rst one is based on ¯lters exchanging discrete signals, while the second
operate on wavelike patterns in a continuum medium. A universal conceptual space
of consciousness would correspond to invariants resulting from the conjoint operation
of both mechanisms. The degree of reproduction of human consciousness would
depend on the arti¯cial mechanisms being similar to those achieved in our evol-
utionary and historical pathways.
Acknowledgments
CNPQ (Brazilian funding agency ��� grant conceded to APJ), Leonid Perlovsky,
David Balduzzi, James Robertson, Chris Nunn, Arnold Trehub, David Rosenthal,
Conceptual Spaces and Consciousness 141
Int.
J. M
ach.
Con
scio
us. 2
011.
03:1
27-1
43. D
ownl
oade
d fr
om w
ww
.wor
ldsc
ient
ific
.com
by 9
3.18
0.53
.211
on
08/1
4/13
. For
per
sona
l use
onl
y.
![Page 16: CONCEPTUAL SPACES AND CONSCIOUSNESS: INTEGRATING COGNITIVE AND AFFECTIVE PROCESSES](https://reader035.vdocuments.us/reader035/viewer/2022073016/5750948b1a28abbf6bb9f086/html5/thumbnails/16.jpg)
Antonio Chella and an anonymous reviewer, for discussion and suggestions that
helped to improve the text and the message.
References
Baars, B. [1997] In the Theater of Consciousness: The Workspace of the Mind (Oxford, NewYork).
Balduzzi, D. and Tononi, G. [2009] \Qualia: The geometry of integrated information," PLoSComputational Biology 5(8), e1000462.
Bechara, A., Dam�asio, A. and Dam�asio, H. [2000] \Emotion, decision making and the orbi-tofrontal cortex," Cerebral Cortex 10, 295�307.
Bennett, M. and Hacker, P. [2005] \Emotion and cortical-subcortical function: Conceptualdevelopments," Progress in Neurobiology 75, 29�52.
Brooks, R., Breazeal, C., Marjanovic, M., Scassellati, B. and Williamson, M. [1998] \The CogProject: Building a humanoid robot," in Computation for Metaphors, Analogy and Agents,Springer Lecture Notes in Arti¯cial Intelligence, pp. 8�13.
Chalmers, D. [1995] \Facing up to the problem of consciousness," Journal of ConsciousnessStudies 2(3), 200�219.
Chalmers, D. [1996] The Conscious Mind (Oxford University Press, New York).Crick, F. and Koch, K. [2003] \A framework for consciousness," Nature Neuroscience 6(2),
119�126.Dam�asio, A. [1994] Descartes' Error. Emotion, Reason, and the Human Brain (Grosset/
Putnam, New York).Dennett, D. C. [1991] Consciousness Explained (Little Brown, Boston).Fell, J. [2004] \Identifying neural correlates of consciousness, the state space approach,"
Consciousness and Cognition 13, 709�729.Frith, C., Perry, R. and Lumer, E. [1999] \The neural correlates of conscious experience, an
experimental framework," Trends in Cognitive Sciences 3(3), 105�114.Gärdenfors, P. [2000] Conceptual Spaces: The Geometry of Thought (MIT Press, Cambridge).Hamero®, S. [2010] \The ‘conscious pilot' ��� dendritic synchrony moves through the brain to
mediate consciousness," Journal of Biological Physics 36, 71�93.He, B. J. and Raichle, M. E. [2009] \The fMRI signal, slow cortical potential and con-
sciousness," Trends in Cognitive Sciences 13, 302�309.Hempel, C. and Oppenheim, P. [1948] \Studies in the logic of explanation," Philosophy of
Science 15, 576�579.Hempel, C. [1965] Aspects of Scienti¯c Explanation, and Other Essays in the Philosophy of
Science (The Free Press, New York).Kay, K. N., Naselaris, T., Prenger, R. J. and Gallant, J. L. [2008] \Identifying natural images
from human brain activity," Nature 452(7185), 352�355.LeDoux, J. [1996] The Emotional Brain (Simon & Schuster Paperbacks, New York).Mandik, P. [1999] \Qualia, space, and control," Philosophical Psychology 12(1), 44�60.Massimini, M., Ferrarelli, F., Murphy, M., Huber, R., Riedner, B., Casarotto, S. and Tononi,
G. [2010] \Cortical reactivity and e®ective connectivity during REM sleep in humans,"Cognitive Neuroscience 1(3), 176�183.
Panksepp, J. [1998] A®ective Neuroscience: The Foundations of Human and Animal Emotions(Oxford University Press, Oxford).
Panksepp, J. [2005] \A®ective consciousness: Core emotional feelings in animals and humans,"Consciousness and Cognition 14, 30�80.
142 A. Pereira Jr. & L. F. Almada
Int.
J. M
ach.
Con
scio
us. 2
011.
03:1
27-1
43. D
ownl
oade
d fr
om w
ww
.wor
ldsc
ient
ific
.com
by 9
3.18
0.53
.211
on
08/1
4/13
. For
per
sona
l use
onl
y.
![Page 17: CONCEPTUAL SPACES AND CONSCIOUSNESS: INTEGRATING COGNITIVE AND AFFECTIVE PROCESSES](https://reader035.vdocuments.us/reader035/viewer/2022073016/5750948b1a28abbf6bb9f086/html5/thumbnails/17.jpg)
Pereira Jr., A. and Furlan, F. A. [2009] \On the role of synchrony for neuron-astrocyteinteractions and perceptual conscious processing," Journal of Biological Physics 35,465�481.
Pereira Jr., A. and Furlan, F. [2010] \Astrocytes and human cognition," Progress inNeurobiology 92, 405�420.
Pereira Jr., A. and Ricke, H. [2009] \What is consciousness? Towards a preliminaryde¯nition," Journal of Consciousness Studies 16(5), 28�45.
Perlovsky, L. [2009] \Toward physics of the mind: Concepts, emotions, consciousness, andsymbols," Physics of Life Review 3(1), 22�55.
Pessoa, L. and Adolphs, R. [2010] \Emotion processing and the amygdala: From a ‘low road' to‘many roads' of evaluating biological signi¯cance," Nature Reviews Neuroscience 11,773�783.
Popper, K. [1963] Conjectures and Refutation: The Growth of Scienti¯c Knowledge (Routledgeand Kegan Paul, London).
Phibram, K. [1991] Brain and Perception Holonomy and Structure in Figural Processing(Lawrence Eribaum, Hillsdale).
Rodriguez, E., George, N., Lachaux, J., Martinerie, J., Renault, B. and Varela, F. J. [1999]\Perception's shadow: Long-distance synchronization of human brain activity," Nature397, 430�433.
Rolls, E. [2000] \Pr�ecis of the brain and emotion," Behavioural and Brain Science 23,177�234.
Rosenthal, D. M. [2002] \Explaining consciousness," in Philosophy of Mind: Classical andContemporaryReadings, Chalmers,D. (ed.) (OxfordUniversityPress,Oxford), pp. 406�421.
Rosenthal, D. M. [2004] \Varieties of higher-order theory," in Higher-Order Theoriesof Consciousness, Rocco, J. Gennaro (ed.) (John Benjamins Publishers), pp. 19�44.
Stanley, R. P. [1999] \Qualia space," Journal of Consciousness Studies 6(1), 49�60.Tononi, G. [2005] \Consciousness, information integration, and the brain," Progress in Brain
Research 150, 109�126.Trehub, A. [2001] The Cognitive Brain (MIT Press, Cambridge).
Conceptual Spaces and Consciousness 143
Int.
J. M
ach.
Con
scio
us. 2
011.
03:1
27-1
43. D
ownl
oade
d fr
om w
ww
.wor
ldsc
ient
ific
.com
by 9
3.18
0.53
.211
on
08/1
4/13
. For
per
sona
l use
onl
y.