Available online at www.sciencedirect.

com

Applied Soft Computing 8 (2008) 1164–1177

www.elsevier.com/locate/asoc

Morphic computing
Germano Resconi a, Masoud Nikravesh b,*
a
Catholic University, Brescia, Italy
b
BISC Program, EECS Department, University of California, Berkeley, CA 94720, USA
Received 21 February 2007; accepted 23 February 2007
Available online 22 October 2007

Abstract
In this paper, we introduce a new type of computation called ‘‘morphic computing’’. Morphic computing is based on field theory and more
specifically morphic fields. Morphic fields were first introduced by Rupert Sheldrake [R. Sheldrake, A New Science of Life: The Hypothesis of
Morphic Resonance (second edition, 1985), Park Street Press; Reprint edition (March 1, 1995), 1981] from his hypothesis of formative causation that
made use of the older notion of morphogenetic fields. Rupert Sheldrake [R. Sheldrake, A New Science of Life: The Hypothesis of Morphic Resonance
(second edition, 1985), Park Street Press; Reprint edition (March 1, 1995), 1981] developed his famous theory, morphic resonance, on the basis of the
work by French philosopher Henri Bergson. Morphic fields and its subset morphogenetic fields have been at the center of controversy for many years in
mainstream science and the hypothesis is not accepted by some scientists who consider it a pseudoscience. We claim that morphic computing is a
natural extension of holographic computation, quantum computation, soft computing, and DNA computing. All natural computations bonded by the
turing machine can be formalised and extended by our new type of computation model—morphic computing. In this paper, we introduce the basis for
this new computing paradigm and its extensions such as quantum logic and entanglement in morphic computing, morphic systems and morphic system
of systems (M-SOS). Its applications to the field of computation by words as an example of the morphic computing, morphogenetic fields in neural
network and morphic computing, morphic fields - concepts and web search, and agents and fuzzy in morphic computing will also be discussed.
# 2007 Elsevier B.V. All rights reserved.

Keywords: Morphic computing; Morphogenetic computing; Morphic fields; Morphogenetic fields; Quantum computing; DNA computing; Soft computing;
Computing with words; Morphic systems; Morphic network; Morphic system of systems

1. Introduction deals with living things. However, morphic fields are more
general than morphogenetic fields and are defined as universal
Inspired by the work of the French philosopher Henri information for both organic (living things) and abstract forms.
Bergson, Rupert Sheldrake (1981) developed his famous theory Sheldrake defined morphic and morphogenetic fields in his
of morphic resonance. Sheldrake’s work on morphic fields, book, The presence of the past [1988] as follows:
based on morphic resonance theory, was published in his well
‘‘The term [Morphic Fields] is more general in its meaning
known book ‘‘A New Science of Life: The Hypothesis of
than Morphogenetic Fields, and includes other kinds of
Morphic Resonance’’ (1981, second edition 1985). Rupert
organizing fields in addition to those of morphogenesis; the
Sheldrake’s [18] theory of Morphic Fields is based on his organizing fields of animal and human behaviour, of social
hypothesis of formative causation that makes use of the older
and cultural systems, and of mental activity can all be
notion of morphogenetic fields. Morphic fields and its subset
regarded as Morphic Fields which contain an inherent
morphogenetic fields have been at the center of controversy for
memory.’’—Sheldrake [19].
many years in mainstream science and the hypothesis is not
accepted by some scientists who consider it a pseudoscience. We claim that morphic fields reshape multidimensional
Morphogenetic fields is a hypothetical biological fields and it space to generate local contexts. For example, the gravitational
has been used by environmental biologists since 1920’s which fields in general relativity are the morphic fields that reshape
space-time space to generate local context where particles
move. Morphic computing reverses the ordinary N input basis
* Corresponding author. 298 HMMB, University of California Berkeley,
Berkeley, CA 94720-1764. Tel.: +1 510 643 4522; fax: +1 510 642 1800.
fields of possible data to one field in the output system. The set
E-mail addresses: [email protected] (G. Resconi), of input fields form a N dimension space or context. The N
[email protected] (M. Nikravesh). dimensional space can be obtained by a deformation of an
1568-4946/$ – see front matter # 2007 Elsevier B.V. All rights reserved.
doi:10.1016/j.asoc.2007.02.018
 G. Resconi, M. Nikravesh / Applied Soft Computing 8 (2008) 1164–1177 1165

Euclidean space and we argue that the morphic fields is the every local invariant property will disappear because every
cause of the deformation. In line with Rupert Sheldrake [18] our system is open and it is not possible to close any local system.
morphic fields is the formative causation of the context. To solve this invariance problem, scientist discovered that the
Now the output field of data is the input field X in morphic local invariant can be conserved with a deformation of the local
computing. to compute the coherence of the X with the context, geometry and the metric of the space. While the form of the
we project X into the context. Our computation is similar to a invariant does not change for the field, the action is changed.
quantum measure where the aim is to find coherence between However, these changes are only in reference to how we write
the behaviour of the particles and the instruments. So the the invariance. In conclusion, we can assume that the action of
quantum measure projects the physical phenomena into the fields can be substituted with deformation of the space. Any
the instrument as a context. The logic of our projection is particle is not under the action of the fields, the invariance as
the same as quantum logic in the quantum measure. In energy, momentum, etc. is true (physical symmetry). However,
conclusion, we compute how a context can implement desired the references that we have chosen change in space and time
results based on morphic computing. and in a way to simulate the action of the field. In this case, all
Our new computation paradigm – morphic computing – is the reference space has been changed and the action of the field
based on field theory and more specifically morphic fields. We is only a virtual phenomena. In this case, we have a different
claim that morphic computing is a natural extension of holo- reference space whose geometry in general is non-Euclidean.
graphic computation, quantum computation, soft computing, and With the quantum phenomena, the problem becomes more
DNA computing. We also claim that all the natural computation complex because the particles are correlated one with one other
bonded by the turing machine can be formalised and extended by in a more hidden way without any physical interaction with
our new type of computation model—morphic computing. fields. This correlation or entanglement generates a structure
In this paper, we will first introduce the basis for our new inside the universe for which the probability to detect a particle
computing paradigm—morphic computing based on field theory. is a virtual or conceptual field that covers the entire Universe.
Then we will introduce its extensions such as quantum logic and
entanglement in morphic computing, morphic systems and 2.2. Morphic computing: basis for quantum, DNA,
morphic system of systems (M-SOS). Then morphic computing’s and soft computing
applications to the field of computation by words will be given.
Finally, we present morphogenetic fields in the neural network Gabor [6] and Fatmi and Resconi [5] discovered the
and morphic computing, morphic field—concepts and web possibility of computing images made by a huge number of
search, and agents and fuzzy in morphic computing. points as output from objects as a set of huge number of points as
input by reference beams or laser (holography). It is also known
2. Morphic computing and field theory: classical and that a set of particles can have a huge number of possible states
modern approach that in classical physics, separate one from the other. However,
only one state (position and velocity of the particles) would be
2.1. Fields possible at a time. With respect to quantum mechanics, one can
have a superposition of all states with all states presented in the
In this paper, we assume that computing is not always related superposition at the same time. It is also very important to note
to symbolic entities as numbers, words or other symbolic that at the same time one cannot separate the states as individual
entities. Fields as entities are more complex than any symbolic entities but consider them as one entity. For this very peculiar
representation of the knowledge. For example, morphic fields property of quantum mechanics, one can change all the
include the universal database for both organic (living) and superpose states at the same time. This type of global
abstract (mental) forms. In classical physics, we represent the computation is the conceptual principle by which we think
interaction among the particles by local forces that are the cause one can build quantum computers. Similar phenomena can be
of the movement of the particles. Also in classical physics, it is used to develop DNA computation where a huge number of DNA
more important to know at any moment the individual values of as a field of DNA elements are transformed (replication) at the
the forces than the structure of the forces. This approach same time and filtered (selection) to solve non-polynomial
considers that the particles are independent from the other problems. In addition, soft-computing or computation by words
particles under the effect of the external forces. But with further extend the classical local definition of true and false value for
development of the theory of particle physics, the researchers logic predicate to a field of degree of true and false inside the
discovered that forces are produced by intermediate entities that space of all possible values of the predicates. In this way, the
are not located in one particular point of space but are at any computational power of soft computing is extended similar to
point of a specific space at the same time. These entities are that which one can find in quantum computing, DNA computing,
called ‘‘fields’’. Based on this new theory, the structure of the and holographic computing. In conclusion, one can expect that
fields is more important than the value itself at any point. In this all the previous approaches and models of computing are
representation of the universe, any particle in any position is examples of a more general computation model called ‘‘morphic
under the effect of the fields. Therefore, the fields are used to computing’’ where ‘‘morphic’’ means ‘‘form’’ and is associated
connect all the particles of the universe in one global entity. with the idea of holism, geometry, field, superposition, globality
However, if any particle is under the effect of the other particles, and so on.
1166 G. Resconi, M. Nikravesh / Applied Soft Computing 8 (2008) 1164–1177

2.3. Morphic computing and conceptual fields—non- Finally, in a deeper form, the biochemical and structural
physical fields morphogenesis is under the control of the gene network
morphogenesis. In fact, any gene will control the activity of the
Morphic computing change or compute non-physical other genes through a continuous adaptive network. Morpho-
conceptual fields. One example is in representing the semantics genesis is essential to brain plasticity so as to obtain
of words. In this case, a field is generated by a word or a homeostasis or the invariant of the fundamental and vital
sentence as sources. For example, in a library the reference functionality of the brain. The predefined vital functions
space would be where the documents are located. At any given interact with the neural network in the brain which activity is
word, we define the field as a map of the position of the oriented to the implementation or projection of the vital
documents in the library and the number of the occurrences function into the neural activities. In conclusion, morphoge-
(values) of the word in the document. The word or source is netic activity is oriented to compensate for the difference
located in one point of the reference space (query) but the field between the vital functions and the neural reply. The field
(answer) can be located in any part of the reference. nature of the designed functionality as input and of the
Complex strings of words (structured query) generate a implemented functionality in the neural network as output
complex field or complex answer by which the structure can be suggest the holistic nature of the neural network activity. The
obtained by the superposition of the fields of the words as neural network with its morphogenesis can be considered as a
sources with different intensity. Any field is a vector in the prototype of morphic computing.
space of the documents. A set of basic fields is a vector space
and form a concept. We break the traditional idea that a concept 2.6. Morphic systems and morphic system of systems (m-sos)
is one word in the conceptual map. Internal structure
(entanglement) of the concept is the relation of dependence We know that system of system or SoS movement study
among the basic fields. The ambiguous word is the source large-scale systems integration. Traditional systems engineer-
(query) of the fuzzy set (field or answer). ing is based on the assumption that if given the requirements the
engineer will give you the system. The emerging system of
2.4. Morphic computing and natural languages—theory of systems context arises when a need or set of needs are met with
generalized constraint a mix of multiple systems, each of which are capable of
independent operation in order to fulfil the global mission or
In a particular case, we know that a key assumption in missions.
computing with words is that the information which is conveyed For example, design optimisation strategies focus on
by a proposition expressed in a natural language or word may be minimizing or maximizing an objective while meeting several
represented as a generalized constraint of the form ‘‘X isr R’’, constraints. These objectives and constraints typically char-
where X is a constrained variable; R is a constraining relation; and acterize the performance of the individual system for a typical
r is an indexing variable whose value defines the way in which R design mission or missions. However, these design strategies
constrains X. Thus, if p is a proposition expressed in a natural rarely address the impact on the performance of a larger system
language, then ‘‘X isr R’’ representing the meaning of p, of systems, nor do they usually address the dynamic, evolving
equivalently, the information conveyed by p. Therefore, the environment in which the system of systems must act. A great
generalised constraint model can be represented by field theory body? of work exists that can address ‘‘organizing’’ a system of
in this way. The meaning of any natural proposition p is given by systems from existing single systems: resource allocation is
the space X of the fields that form a concept in the reference space currently used in any number of fields of engineering and
or objective space, and by a field R in the same reference. We note business to improve the profit? other systems optimisation. One
that a concept is not only a word, but is a domain or context X reason for this new emphasis on large-scale systems is that
where the propositions p represented by the field R is located. The customers want solutions to provide a set of capabilities, not a
word in the new image is not a passive entity but is an active single specific vehicle or system to meet an exact set of
entity. In fact, the word is the source of the field. We can also use specifications. Systems engineering is based on the assumption
the idea that the word as an abstract entity is a query and the field that if given the requirements the engineer will give you the
as set of instances of the query is the answer. system ‘‘There is growing recognition that one does not
necessarily attack design problems solely at the component
2.5. Morphogenetic and neural network level.’’ The cardinal point for SoS studies is that the solutions
are unlikely to be found in any one field There is an additional
A neural network is a complex structure that connects simple consideration: A fundamental characteristic of the problem
entities denoted as neurons. The main feature of the neural areas where we detect SoS phenomena will be that they are
network is the continuous evolution in complexity of the open systems without fixed and stable boundaries and adapt to
interconnections. This evolutionary process is called morpho- changing circumstances. In SoS there is common language and
genesis. Besides the evolution in the structure, we also have an shared goals, communications helps form communities around
evolution in the biochemical network inside the neurons and the a confluence of issues, goals, and actions. Therefore, if we wish
relations among neurons. The biochemical morphogenesis is to increase the dimensionality we need tools not to replace
useful for adapting the neurons to the desired functionality. human reasoning but to assist and support this type of
 G. Resconi, M. Nikravesh / Applied Soft Computing 8 (2008) 1164–1177 1167

associative reasoning. Important effects occur at multiple for finding appropriate searches to obtain the relevant answer.
scales, involving not only multiple phenomena but phenomena For the agent interpretation of the fuzzy set, the net of the Web
of very different types. These in turn, are intimately bound to is structured as a set of conflicting and in many case irrational
diverse communities of effected individuals. Human activity agents whose task is to create any concept. Agents produce
now yields effects on a global scale. Yet we lack the tools to actions to create answers for ambiguous words in the Web. A
understand the implications of our choices. Given a set of structured query in RDF can be represented as a graph of three
prototype systems, integrations (superposition) of these elementary concepts as subject, predicate and complement in a
systems generate a family of meta-systems. Given the conceptual map. Every word and relationship in the conceptual
requirement X the MS (morphic system) reshape in a minimum map are variables whose values are fields which their
way or zero the X. The reshaped requirements belong to the superposition gives the answer to the query. Because we are
family of meta-systems and can be obtained by an integration of more interested in the meaning of the query than how we write
the prototype systems. In conclusion, the MS is the instrument the query itself, we are more interested in the field than how we
to implement the SoS. To use MS we represent any individual produce the field by the query. In fact, different linguistic
system as a field the superposition with the field of the other representations of the query can give the same field or answer.
systems is the integration process. This is similar to the In the construction of the query, we use words as sources of
quantum computer where any particle (system) in quantum fields with different intensities. With superposition, we obtain
physics is a field of uncertainty (density of probability). The the answer for our structured query. We structure the text or
integration among particles is obtained by a superposition of query to build the described field or meaning. It is also possible
the different uncertainties. The meta-system or SoS is the result to use the answer, as a field, to generate the intensity of the words
of the integration. Any instrument in quantum measure has only as sources inside a structured query. The first process is denoted
a few possible numbers of integration. So the quantum measure READ process by which we can read the answer (meaning) of
reshape the given requirement or physical integration in a way the structured query. The second process is the WRITE process
to be one of the possible integrations in the instrument. So the by which we give the intensity or rank of the words in a query
quantum instrument is a particular case of the SoS and of the when we know the answer. As an analogy to holography, the
MS. In MS we also give suitable geometric structure (space of WRITE process is the construction of the hologram when we
the fields of the prototype systems) to represent integration in know the light field of the object. The READ is the construction
SoS and correlation among the prototype systems. With this of the light field image by the hologram. In the holography, the
geometry we give an? invariance property for transformation of READ process uses a beam of coherent light as a laser to obtain
the prototype systems and the requirement X. So we adjoin to the image. Now in our structured query, the words inside of text
the SoS integration a dynamical process which dynamical law are activated at the same time. The words as sources are coherent
is the invariant. Since the geometry is in general a non- in the construction by superposition of the desired answer or
Euclidean geometry, we can define for the SoS a morpho field field. Now the field image of the computation by words in a crisp
(MF) which action is to give the deformation of the geometry and fuzzy interpretation prepares the implementation of the
from Eucliden (independent prototype system) to a non- morphic computing approach to the computation by words. In
Euclidean (dependent prototype system). The morpic field is this way, we have presented an example of the meaning of the
comparable to the gravity field in general relativity. So the new type of computation, ‘‘morphic computing’’.
model of quantum mechanics and general relativity are a
special case of MS and are also the physical model for the SoS. 2.8. Morphic computing: basic concepts

2.7. Morphic computing and agents—non-classical logic Morphic computing is based on the following concepts:

In the agent image, where only one word (query) as a source (1) The concept of field in the reference space.
is used for any agent, the field generated by the word (answer) is (2) The fields as points or vectors in the N dimension Euclidean
a Boolean field (the values in any points are true or false). space of the objects (points).
Therefore, we can compose the words by logic operations to (3) A set of M  N basis fields in the N dimensional space. The
create complex Boolean expression or complex Boolean query. set of M fields are vectors in the N dimensional space. The
This query generates a Boolean field for any agent. This set of set of M vectors form a non-Euclidean subspace H (context)
agents creates a set of elementary Boolean fields whose of the space N. The coordinates Sa in M of the field X are the
superposition is the fuzzy set represented by a field with fuzzy contro-variant components of the field X. The components
values. The field is the answer to the ambiguous structured of X in M are also the intensity of the sources of the basis
query whose source is the complex expression p. The fields with field. The superposition of the basis field with different
fuzzy values for complex logic expression are coherent with intensity give us the projection Q of X or Y = QX into the
traditional fuzzy logic with a more conceptual transparency space H When M < N the projection operator of X into H
because it is found on agents and Boolean logic structure. As define a constrain or relation among the components of Y.
pointed out in Ref. [16] the Web is a large unstructured and in (4) With the tensor calculus with the components Sa of the
many cases conflicting set of data. So in the World Wide Web, vector X or the components of more complex entity as
fuzzy logic and fuzzy sets are essential parts of a query and also tensors, we can generate invariants for any unitary
1168 G. Resconi, M. Nikravesh / Applied Soft Computing 8 (2008) 1164–1177

transformation of the object space or the change of the that interferes with the physical phenomena as H interferes with
basis fields. the input field X.
(5) Given two projection operators Q1, Q2 on two spaces H1, H2 A deeper connection exists between the projection operator
with dimension M1 and M2 we can generate the M = M1M2, lattice that represents the quantum logic and morphic
with the product of Y1 and Y2 or Y = Y1Y2. Any projection Q computing processes (see Eddie Oshins [13]).
into the space H or Y = QX of the product of the basis fields Because any fuzzy set is a scalar field of the membership
generate Y. When Y 6¼ Y1Y2 the output Y is in entanglement values on the factors (reference space) (Wang and Sugeno [23]).
state and cannot separate in the two projections Q1 and Q2. We remember that any concept can be viewed as a fuzzy set in
(6) The logic of the morphic computing entity is the logic of the the factor space. So at the fuzzy set we can introduce all the
projection operators that is isomorphic to the quantum logic. processes and concepts that we utilise in morphic computing.
For the relation between concept and field, we introduce in
The information can be coded inside the basis fields by the the field theory an intrinsic fuzzy logic. So in morphic
relation among the basis fields. In morphic computing, the computing, we have an external logic of the projection or
relation is represented by a non-Euclidean geometry which measure (quantum logic) and a possible internal fuzzy logic of
metric or expression of the distance between two points shows the fuzzy interpretation of the fields.
this relation. In the end, because we also use agents’ superposition to
The projection operator is similar to the measure in quantum define fuzzy sets and fuzzy rules, we can again use morphic
mechanics. The projection operator can introduce constrains computing to compute the agents inconsistency and irration-
inside the components of Y. ality. Thus, fuzzy set and fuzzy logic are part of the more
The sources are the instrument to control the image Y in general computation denoted as morphic computing.
morphic computing. There is a strong analogy between morphic
computing and computation by holography and computation by 3. Reference space, space of the objects, space of the
secondary sources (Jessel) in the physical field. fields in the morphic computing
The computation of Y by X and the projection operator Q that
project X into the space H give his result when the Y is similar to Given the n dimensional reference space (R1, R2, . . ., Rn),
X. In this case, the sources S are the solution of the computation. any point. P = (R1, R2, . . ., Rn) is an object.
We see the analogy with the neural network where the solution Now we create the space of the objects which dimension is
is to find the weights wk at the synapse. In this paper, we show equal to the number of the points and the value of the
that the weights are sources in morphic computing. coordinates in this space is equal to the value of the field in the
Now, it is possible to compose different projection operators point. We call the space of the points ‘‘space of the objects’’.
in a network of morphic systems. It is obvious to consider this Any field connect all the points in the reference space and
system as a system of systems. is represented as a point in the object space. The components
Any morphic computation is always context dependent of this vector are the value of the field in the different points.
where the context is H. The context H by the operator Q define a We know that each of the two points connected by a link
set of rules that are comparable with the rules implemented in a assume the value of one of the connections. All the other
digital computer. So when we change the context with the same points assume zero value. Now any value of the field in a point
operations, we obtain different results. We can control the can be considered as a degree of connection of this point with
context in a way to obtain wanted results. When any projection all the others. Therefore, in one point where the field is zero,
operator of X or QX is denoted as a measure, in analogy with we can consider this point as non-connected to the others. In
quantum mechanics, any projection operator depends on the fact, because the field in this point is zero the other points
previous projection operator. In the measure analogy, any cannot be connected by the field to the given point. In
measure depends on the previous measures. So any measure is conclusion, we consider the field as a global connector of the
dependent on the path of measures or projection operators that objects in the reference space. Now inside the space of the
we realise before or through the history. So we can say that objects, we can locate any type of field as vectors or points. In
different projection operators are a story (see Roland Omnès field theory, we assume that any complex field can be
[12] in quantum mechanics stories). considered as a superposition of prototype fields whose model
The analogy of the measure also gives us another intuitive is well known.
idea of morph computing. Any measure become a good The prototype fields are vectors in the space of the objects
measure when gives us an image Y of the real phenomena X that that form a new reference or field space. In general, the field
is similar, when the internal rules to X are not destroyed. In the space is a non-Euclidean space. In conclusion, any complex
measure process, the measure is a good measure. The same for field Y can be written in this way
morphic computing, the computation is a good computation
when the projection operator does not destroy the internal Y ¼ S1 H 1 ðR1 ; . . . ; Rn Þ þ S2 H 2 ðR1 ; . . . ; Rn Þ þ   
relation of the field in input X. þ Sn H n ðR1 ; . . . ; Rn Þ ¼ HðRÞS (1)
The analogy with the measure in quantum mechanics is also
useful to explain the concept of morphic computing because the In Eq. (1), H1, H2, . . ., Hn are the basic fields or prototype
instrument in the quantum measure is the fundamental context fields and S1, S2, . . ., Sn are the weights or source values of the
 G. Resconi, M. Nikravesh / Applied Soft Computing 8 (2008) 1164–1177 1169

Fig. 1. Two different basic fields in the two dimensional reference space (x, y).

Fig. 2. Example of superposition of elementary fields F1, F2.

basic fields. We assume that any basic field is generated by a The values in Table 1 is represented by the following matrices
source. The intensity of the prototype fields is proportional to 2 3 2 3
the intensity of the sources that generates the field itself. H 1;1 H 1;2    H 1;N X1
6 H 2;1 H 2;2    H 2;N 7 6 X2 7
H¼64 
7;
5 X¼6 7
4  5
3.1. Example of the basic field and sources   
H M;1 H M;2    H M;N XM
In Fig. 1, we show an example of two different basic fields in
The matrix H is the relation among the prototype fields F k
a two dimensional reference space (x, y). The general equation
and the points Ph. At this point, we are interested in the
of the fields is
computation of the sources S by which they give the best linear
2
þðyy0 Þ2 Þ model of X by the elementary field values. Therefore, we have
Fðx; yÞ ¼ S½ehððxx0 Þ  (2) the superposition expression
the parameters of the field F 1 are S = 1, h = 2 and x0 = 0.5 and 2 3 2 3 2 3
H 1;1 H 1;2 H 1;n
y0 = 0.5, the parameters of the field F 2 are S = 1, h = 2 and 6H 7 6H 7 6H 7
x0 = 0.5 and y0 = 0.5. 6 2;1 7 6 2;2 7 6 2;n 7
Y ¼ S1 6 7 þ S2 6 7 þ    þ Sn 6 7 ¼ HS (3)
For the sources S1 = 1 and S2 = 1 the superposition field F 4  5 4  5 4  5
that is shown in Fig. 2 is F = F 1 + F 2. For the sources S1 = 1 H M;1 H M;2 H M;n
and S2 = 2, the superposition field F that is shown again in
Table 1
Fig. 2 is F = F 1 + 2F 2. Fields values for M points in the reference space

3.2. Computation of the sources H1 H2  HN Input Field X

P1 H1,1 F1,2  H1,N X1
To compute the sources Sk, we represent the prototype field P2 H2,1 H2,2  H2,N X2
     
Hk and the input field X in a Table 1 where the objects are the
PM HM,1 HM,2  HM,N XM
points and the attribute are the fields.
1170 G. Resconi, M. Nikravesh / Applied Soft Computing 8 (2008) 1164–1177

Fig. 3. Elementary morphic computing.

Then, we compute the best sources S in a way the difference Fig. 5. The fields H1 and H2 are the space of the fields. The coordinates of the
jY  Xj is the minimum distance for any possible choice of vectors H1 and H2 are the values of the fields in the three points P1, P2, P3.
the set of sources. It is easy to show that the best sources are
obtained by the expression
With the property Q2X = QX.
1 Therefore, the input X can be separated in two parts
S ¼ ðH T HÞ H T X (4)
X ¼ QX þ F
Given the previous discussion and field presentation, the
elementary Morphic Computing element is given by the input– where the vector F is perpendicular to the space H as we can see
output system as shown in Fig. 3. in a simple example given in Fig. 6.
Fig. 4 shows network of elementary morphic computing Now, we try to extend the expression of the sources in the
with three set of prototype fields and three type of sources with following way.
one general field X in input and one general field Y in output and Given G(G) = GTG and G(H) = HTH and S* = [G(H) +
intermediary fields from X and Y. G(G)1HTX and [G(H)+G(G)]S* = HTX. So far S* =
When H is a square matrix, we have Y = X and (HTH)1HTX + Qa = Sa + Va we have

S ¼ H 1 X and Y ¼ X ¼ HS (5) ½GðHÞ þ GðG ÞðSa þ Va Þ ¼ H T X;

1
ðH T HÞðH T HÞ H T X þ ½GðHÞ þ GðG ÞVa ¼ H T X;
Now for any elementary computation in the morphic
computing, we have the following three fundamental spaces. GðG ÞSa þ ½GðHÞ þ GðG ÞVa ¼ 0
(1) The reference space.
and
(2) The space of the objects (points).
(3) The space of the prototype fields. 1
S ¼ S þ V ¼ ðH T H þ G T G Þ H T X
Fig. 5 shows a very simple geometric example when the where for V is function of S by the equation
number of the objects are three (P1, P2, P3) and the number of GðG ÞS þ ½GðHÞ þ GðG ÞV ¼ 0
the prototype fields are two (H1, H2). The space which
coordinates are the two fields is the space of the fields. For non-square matrix and/or singular matrix, we can use the
Note that the output Y = HS is the projection of X into the generalized model given by Nikravesh as follows:
space H 1
S ¼ ðH T LT LH Þ H T LT LX
1
Y ¼ HðH T HÞ H T X ¼ QX 1
¼ ððLHÞT ðLHÞ Þ ðLHÞT LX
where we transform by L the input and the references H.
The value of the variable D (metric of the space of the field)
is computed by the expression (6)
D2 ¼ ðHSÞT ðHSÞ ¼ ST H T HS ¼ ST GS ¼ ðQXÞT QX (6)

Fig. 6. Projection operator Q and output of the elementary morphic computing.

Fig. 4. Shows the network of morphic computing. We see that X = QX + F, where the sum is the vector sum.
 G. Resconi, M. Nikravesh / Applied Soft Computing 8 (2008) 1164–1177 1171

For the unitary transformation U for which, we have UTU = I

and H0 = UH the prototype fields change in the following
way:H 0 ¼ UH; G0 ¼ ðUHÞT ðUHÞ ¼ H T U T UH ¼ H T Ha-
and
1
S0 ¼ ½ðUHÞT ðUHÞ ðUHÞT Z ¼ G1 H T U T Z
¼ G1 H T ðU 1 ZÞ
Fig. 7. Expressions for projection operator that from the morphic computing
For Z = UX we have S0 = S and the variable D is invariant, for entity.
the unitary transformation U.
We remark that G = HT H is a quadratic matrix that gives the
metric tensor of the space of the fields. When G is a diagonal Now we give an example of the projection logic and lattice
matrix the entire elementary field are independent one from the in this way:
other. But when G has non-diagonal elements, in this case Given the elementary field references
the elementary fields are dependent on from the other. Among 2 3
1
the elementary fields there is a correlation or a relationship and     p ffiffi
ffi
1 0 6 27
the geometry of the space of the fields is a non-Euclidean H1 ¼ ; H2 ¼ ; H3 ¼ 64 1 5
7
0 1
geometry. pffiffiffi
2
4. Quantum logic and entanglement in morphic For which we have the projection operators
computing  
T 1 T 1 0
Q1 ¼ H 1 ðH1 H 1 Þ H1 ¼ ;
In the morphic computing, we can make computation on the 0 0
 
context H as we make the computation on the Hilbert space. Now 1 0 0
Q2 ¼ H 2 ðH2T H 2 Þ H2T ¼ ;
we have the algebras among the context or spaces H. in fact we 0 1
have H ¼ H 1 H 2 where  is the direct sum. For example 21 13
given H 1 ¼ ðh1;1 h1;2 ; . . . ; h1; p Þ where h1,k are the basis fields in 6 7
Q3 ¼ H 3 ðH3T H 3 Þ H3T ¼ 4 2 2 5
1
H1. H 2 ¼ ðh2;1 ; h2;2 ; . . . ; h2;q Þ where h2,k are the basis fields in 1 1
H2. 2 2
So we have H ¼ H 1 H 2 ¼ ðh1;1 ; h1;2 ; . . . ; h1; p ; h2;1 ; h2;2 ;
. . . ; h1;q Þ. The intersection among the context is H ¼ H 1 \ H 2 . With the lattice logic we have
The space H is the subspace in common to H1, H2. In fact for  
1 0
the set V1 and V2 of the vectors H 1;2 ¼ H 1 H 2 ¼ ; Q1 _ Q2
0 1
 
V 1 ¼ S1;1 h1;1 þ S1;2 h1;2 þ    þ S1; p h1; p ; T 1 1 0
¼ H 1;2 ðH1;2 H 1;2 Þ H 1;2 ¼ ; H 1;3 ¼ H 1 H 3
V 2 ¼ S2;1 h2;1 þ S2;2 h2;2 þ    þ S2;q h2;q 0 1
2 3
1
The space or context H = H1 \ H2 include all the vectors in
6 1 pffiffi2ffi 7 1
V1 \ V2. ¼64
7 T
1 5; Q1 _ Q3 ¼ H 1;3 ðH1;3 H 1;3 Þ H 1;3
Given a space H, we can also built the orthogonal space H? 0 pffiffiffi
of the vectors that are all orthogonal to any vectors in H. 2
 
Now we have this logic structure 1 0
¼ ; H 2;3 ¼ H 2 H 3
0 1
QðH 1 H 2 Þ ¼ Q1 _ Q2 ¼ Q1 OR Q2 2 3
1
where Q1 is the projection operator on the context H1 and Q2 is 6 0 p ffiffi
ffi
¼6 27 7 T 1
the projection operator on the context H2. 4 1 5; Q2 _ Q3 ¼ H 2;3 ðH2;3 H 2;3 Þ H 2;
1 pffiffiffi
QðH 1 \ H 2 Þ ¼ Q1 ^ Q2 ¼ Q1 AND Q2 ; 2
 
QðH ? Þ ¼ Q  ID ¼ : Q ¼ NOT Q 1 0
¼
0 1
In fact, we know that QX  X = (Q  ID)X is orthogonal to
Y and so orthogonal to H. In this case the operator (Q  ID) is and
 
the not operator. Now it easy to show [9] that the logic of the 0
projection operator is isomorphic to the quantum logic and H ¼ H1 \ H2 ¼ H1 \ H3 ¼ H2 \ H3 ¼ and Q1 ^ Q2
0
form the operator lattice for which the distributive low  
(interference) is not true. In Fig. 7, we show an expression in the 0 0
¼ Q1 ^ Q3 ¼ Q2 ^ Q3 ¼
projection lattice for the morphic computing. 0 0
1172 G. Resconi, M. Nikravesh / Applied Soft Computing 8 (2008) 1164–1177

and in conclusion we have the lattice fact Q1 ^ Q3 project in the zero point. A union of the zero point
cannot create the two dimensional space. the non-distributive
property assume that among the projection operators there is an
entanglement or relation that we destroy when we separate the
operators one from the other.
Given two references or contexts H1, H2 the tensor product
H = H1  H2 is the composition of the two independent
contexts in one.
We can prove that the projection operator of the tensor
We remark that product H is the tensor product of Q1, Q2. So we
1
haveQ ¼ HðH T HÞ H T ¼ Q1 Q2
ðQ1 _ Q2 Þ ^ Q3 ¼ Q3 but The sources are Sab = Sa1 Sb2 . So we have
ðQ1 ^ Q3 Þ _ ðQ2 ^ Q3 Þ ¼ 0 _ 0 ¼ 0
Y ¼ Y 1 Y 2 ¼ ðH 1 H 2 ÞSab
When we try to separate Q1 from Q2 in the second
expression the result change. Between Q1 and Q2 we have a The two morphic system are independent from one another.
connection or relation (Q1 and Q2 generate the two dimensional The output is the product of the two outputs for the any morphic
space) that we destroy when we separate one from the other. In system. Now we give some examples

2 2 3 21 33
1
1 1
6 62 7 62 77
6 6 7 6 77
6 6 17 16 177
6 16 0 7 6 0 77
6 6 27 26
27 7
6 4 5 4 57
6 1 1 1 1 7
6 7
6 2 2 7
2 3 2 3 2 3 6 22 2 3 2 3 7
1 1 1 6 1 1 7
1 1 1H 2 H2 6 1 1 7
6 27 62 7 6 2 6
7 6 62 7 62 77
6 7 6 7 6 7 616 7 6 77
61 7 6 17 61 7 6 17 6 177
H1 ¼ 6 1 7; H 2 ¼ 6 0 7; H 1 H 2 ¼ H 1 a;b H 2 g;d ¼ 6 H 2 0H 2 7 ¼ 6 6 60 7 16 0 77 ¼ H a;b;g;d
62 7 6 2 7 62 7 626 2 7 6 2 777
4 5 4 5 4 5 6 4 5 4 5
1 1 1 1 6 1 1 1 1 7 7
1 1H 2 H2 6
2 2 2 2 6 2 2 2 2 7 7
6 21 3 21 37
6 1 1 7
6 6 7 62 77
6 62 7 6 77
6 6 177
6 16 0 1 7 7 260 16
77
6 6 77
6 4 275
6
4 2 57
4 1 1 1 1 5
2 2 2 2
At every value of the basis fields we associate the basis field H2 multiply for the value of the basis fields in H1. Now because we have
22 1 1 3
2 3
1 1
0 63 3 3 7
62 27 6 7
T 1 T 6 7 T 1 T 61 2 17
Q1 ¼ H 1 ðH1 H 1 Þ H1 ¼ 6 0 1 0 7; Q2 ¼ H 2 ðH2 H 2 Þ H2 ¼ 6  7;
41 15 63 3 37
4 5
0 1 1 2
2 2
2 3 3 3 3
1 1 2 3 2 3 21 3
1 1
6 2 Q2 0Q2 2 Q2 7 637 647
X
63 27
6 7 6 7 6 7 6 7
6
Q ¼ Q1 Q2 ¼ 6 0Q2 1Q2 0Q2 7; 7 X 1 ¼ 6 0 7 ;X 2 ¼ 6 0 7 ;X ¼ X 1 X 2 ¼ 6 0X 2 7;
41 5 415 415 4 5
1 1
Q2 0Q2 Q2 X2
2 2 3 4 3
2 4 3 2 33

6 7 6 7
S1 ¼ ðH1T H 1 Þ H1T X 1 ¼ 4 9 5 ;S2 ¼ ðH2T H 2 Þ H2T X 1 ¼ 4 2 5
1 1
2 4

9 3
 G. Resconi, M. Nikravesh / Applied Soft Computing 8 (2008) 1164–1177 1173

and
2 3
4
6 S 2 7
Sab ¼ S1 S2 ¼ 4 92 5
 S2
9
for Fig. 8. Tensor product for independent projection operators or measures.
2 7 3
2 3
1 Table 2
6 12 7
637 6 7 Term (word), document and complex text G
6 7 6 2 7
Y 1 ¼ H 1 S1 ¼ 6 0 7 ; Y 2 ¼ H 2 S2 ¼ 6 7; Word1 Word2  WordN Concept X
4 5 6 3 7
1 4 5
1 Document1 K1,1 K1,2  K1,N X1
3  Document2 K2,1 K2,2  K2,N X2
12
21 3      
Y DocumentM KM,1 KM,2  KM,N XM
63 27
6 7
Y ¼ Y 1 Y 2 ¼ 6 0Y 2 7
4 5 found the suitable sources S by which we can project X into the
1
Y2 knowledge inside the library H.
3
The library can be represented in a geometric as shown in
In conclusion, we can say that the computation of Q, S and Y Fig. 10.
by H and X can be obtained only with the results of Q1, S1, Y1 Recently much work has been done to realise conceptual
and Q2, S2, Y2 independently. Now when H and X cannot be maps of lexicon for natural language. This work has a foundation
write as a scalar product of H1, X1 and H2, X2, the context H and in description logic. With a conceptual map of natural language
the input X are not separable in the more simple entities so are in and description logic, we can create a semantic web. The problem
entanglement. with a semantic web is the integration of different concepts in a
In conclusion with the tensor product, we can know if two geometric frame. We are in the same situation as the quantum
measures or projection operators are dependent or independent computer where the geometric Hilber space is the geometric tool
from one another. to show the global properties of the classical logic and quantum
So we have for the tensor product of the context that all the logic. We propose in this paper to give a geometric image of
other elements of the entity are obtained by the tensor product knowledge. The meaning of the words and concepts are not
(Fig. 8). located in the axiomatic description logic, in the conceptual map
or in the lexical structure. The meaning is inside the space of the
5. Field theory, concepts and web search fields of instances of the words or documents. A prototype set of
concepts, inside the space of the documents as object, generate an
To search in the web, we use the term-document matrix to infinite number of other concepts in a space. Any concept inside
obtain the information retrieved. In Table 2, we show the data the space of the concepts is obtained by an integration process of
useful in obtaining the desired information in the Web. the prototype concepts. Now any new concept can be compared
Where Ki,j is the value of the wordj in the documenti. The word with the concepts that belong to the space of the concepts. The
in Table 2 is a source of a field which values are the values in the MS can reshape, with the minimum change the new concept X
position space of the documents. Any document is one of the into another concept Y. Concept Y belongs to the set of concepts
possible positions of the word. In a Web search, it is useful to obtained by the integration of the prototype concepts. Y is inside
denote the words and complex text in a symbolic form as queries,
the answers are the fields generated by the words or text as
sources. Any ontology map is a conceptual graph in RDF
language where we structure the query as a structured variable.
The conceptual map in Fig. 9 is the input X of a complex field
obtained by the superposition of the individual words in the map.
With the table two we have that the library H is the main
reference for which we have
2 3
K 1;1 K 1;2    K 1;N
6 K 2;1 K 2;2    K 2;N 7
H¼6 4 
7
   5
2 3
K M;1 K M;2    K M;N X1
6 X2 7
where M > N.For any new concept in input X ¼ 6 7
4 ::: 5 we can
Fig. 9. Conceptual map as structured query. The map is a structured variable
XM
whose answer or meaning is the field G in the documents space located in the Web.
1174 G. Resconi, M. Nikravesh / Applied Soft Computing 8 (2008) 1164–1177

The weights for the neural network can be obtained by

different methods as back propagation or others.
Now we study the same Boolean function by the morphic
computing network or system of system. In fact at the begin we
consider the simple morphic system with
2 3 2 3 2 3
0 0 0 0
607 617 60 17
6 7 6 7 6 7
H ¼ X[Y ¼ 6 7[6 7 ¼ 6 7;
415 405 41 05
Fig. 10. Geometric representation of the library (documents) and concepts
1
1 1 1
(words). 2 3
0
the space of the prototype fields that represent prototype 617
6 7
concepts. input ¼ Z ¼ XY ¼ 6 7
415
MS can compute the part of X that cannot be integrated in the
space of the concepts. Transformation of the space of the 0
concepts generate a set of new prototype concepts by which we where the input of the morphic system is the field of all possible
can represent the dynamical process of the concepts and the values for Z. The field reference H or the context is made by
possible conceptual invariance. In this way a huge set of concepts the fields one are all possible values for the input X and the other
can be reduced to a more simple set of prototype concepts and are all possible values for Y.
invariance. In the semantic web based on description logic and With the context H (field references) and the input Z we
the conceptual map, we cannot reduce the concepts to a simple compute the sources and the output. So we have
set of basic concepts which. With the description of invariance 213
conceptual frame and the type of geometry (metric, Euclidean
6 7
S ¼ ðH T HÞ H T input ¼ 4 3 5;
and non-Eucliden) we believe we can improve the primitive 1

semantic structure base on the lexical description of the concepts. 1

3
2 3
6. Morphogenetic field in neural network and morphic 0
2 3 2 3
computing: morphic systems and morphic system of 0 0 617
6 7
systems (m-sos) 16 7 6 7 637
607 1617 6 7
output ¼ HS ¼ 6 7 þ 6 7 ¼ 6 1 7
3415 3405 6 7
We know that in traditional neural network one of the 637
1 1 4 5
problems was the realisation of the Boolean function XOR. The 2
solution of this problem was realised by the well-known back 3
propagation. Now the physiological realisation in the brain of We remark that the space generate by the fields X and Y in H is
the Boolean functions are obtained by the genetic change on the given by the superposition
synaptic form. The morphogenetic field is the biochemical 2 3 2 3 2 3
system that change the synaptic structure. The solution of the 0 0 0
XOR problem and other similar problems is relate to the hidden 607 6 1 7 6 S2 7
output ¼ S1 6 7 6 7 6
4 1 5 þ S2 4 0 5 ¼ 4 S1 5
7
layer of neurons which activity is not used directly to the output
computation but are an intermediary computation useful to 1 1 S1 þ S2
obtain the wanted output. In Fig. 11, we show the neural Now we know that the Q projection operator for
network that realise the function
2 3 2 3 2 3
Z ¼ XX OR Y ¼ XY 0 0 0 0
607 617 60 17
H ¼ X[Y ¼ 6 7 6 7 6
415[405 ¼ 41
7 (7)
For the previous neural network we have 05
  1 1 1 1
2 2 2
Z ¼ XY ¼ F Xþ Y  ðX þ YÞ is
3 3 3
2 3
0 0 0 0
6 2 1 17
60  7
6 3 3 37
1 6 7
Q ¼ HðH T HÞ H T ¼ 6 1 2 17
60  7
6 3 3 37
4 5
1 1 2
0
3 3 3
Fig. 11. Neural network for the Boolean function XOR.
 G. Resconi, M. Nikravesh / Applied Soft Computing 8 (2008) 1164–1177 1175

Because the determinant of Q is equal to zero, the colon 2 3

0 0 0
vectors are not independent so for 60 1 17
2 2 Now for the reference or context H ¼ 6 7
13 213 41 0 15

6 3 37 637 1 1 1
6 7 6 7 2 3
6 1 2 7 617
H ¼ 6 7; X¼6 7 0 1 1
6 3 3 7 637 That we can reduce to H ¼ 4 1 0 1 5. Now because H
4 5 4 5
1 1 2
1 1 1
3 3 3 is a quadratic matrix, in the projection we have no limitation or
we have constrain so when in input we have
  2 3
T 1 1
T 0
S ¼ ðH HÞ H X ¼ and y 4 ¼ S 1 y 2 þ S2 y 3 617
1 input ¼ Z ¼ XY ¼ 6 7
415
¼ y2 þ y3 (8) 0
So we have the constrain by the projection operator that any By the projection operator Q we have that the output = input.
Y must have this form In this case we can compute the sources S in a traditional way
2 3 for the context
0
6 y2 7 2 3
Y¼6 7 (9) 0 1 1
4 y3 5
H ¼ 41 0 15
y2 þ y3 1 1 1
Now because the output that we want must have this form The sources are
y2 > 0; y3 > 0; y2 þ y3 < minðy2 ; y3 Þ 2
3
1
1
But this is impossible so the projection operator and S ¼ ðH T HÞ H T X ¼ 4 1 5
reference H that we use put a limitation or a constraint that the 2
XOR Boolean function cannot satisfy. Now given the reference (4) we have the constrain (5). Now
for the input
2 3
0
617
input ¼ 6415
7 (10)
1
The output is (6) that is coherent or similar to the input. In
fact we have that for
2 3
0
6 y2 7
Y¼6 4 y3 5
7

y2 þ y3
because in input we have the property. x1 = 0, x2 > 0, x3 > 0,
x4 > 0 and in output we have the same property y1 = 0, y2 > 0,
y3 > 0, y4 = y2 + y3 > 0.
For the context (4) and the input (7) the sources are
2 3
2
T 1 T 637
S ¼ ðH HÞ H X ¼ 4 2 5
3
In conclusion, we found the same weights as in Fig. 12 with
the morphic system.
We can conclude that in the ordinary back propagation we
compute the weights in neuron network by a recursive
correction of the errors in output. In morphic computing we
Fig. 12. Network of morphic systems or system of systems. are interested in the similarity between the input and output. So
1176 G. Resconi, M. Nikravesh / Applied Soft Computing 8 (2008) 1164–1177

we are interested more in the morphos (form or structure) that in In fact we have X k Y k þ X k Y k ¼ Y k and X k Y k ¼ Y k  X k Y k
the individual values in input and output. in the same way we have

7. Agents and fuzzy in morphic computing X k þ Y k ¼ X k þ X k Y k þ X k Y k ¼ X k ð1 þ Y k Þ þ X k Y k

¼ Xk þ Xk Y k
Given the word ‘‘warm’’ and three agents that are named
Carlo, Anna and Antonio, we assume that any agent is a source because in the fuzzy set for the Zadeh rule we have
for a Boolean fields in Fig. 13.
The superposition of the three Boolean fields in Fig. 13 is the m½ðX 1 ; . . . ; X p ÞðX 1 ; . . . ; X p Þ
fuzzy set ¼ min½mðX 1 ; . . . ; X p Þ; mðX 1 ; . . . ; X p Þ > 0;
mðA; BÞ ¼ S1 F 1 ðA; BÞ þ S2 F 2 ðA; BÞ þ S3 F 3 ðA; BÞ (11) m½ðX 1 ; . . . ; X p ÞðX 1 ; . . . ; X p Þ

At the word ‘‘warm’’ that is the query we associate the ¼ max½mðX 1 ; . . . ; X p Þ; mðX 1 ; . . . ; X p Þ < 1
answer m(A, B) that is a field. With the superposition process, the negation in (11) is not compatible with the other operations.
we can introduce the logic of the fuzzy fields in this way. So in the decomposition of the vector
When the logic variables values for p agents assume the
usual 1 and 0 (Boolean fields), we have the operations ðX 1 ; X 2 ; . . . ; X p Þ ¼ ðFðX 1 Þ; FðX 2 Þ; . . . ; FðX p ÞÞ (15)

ðX 1 ; X 2 ; . . . ; X p ÞðY 1 ; Y 2 ; . . . ; Y p Þ ¼ ðY 1 X 1 ; Y 2 X 2 ; . . . ; Y p X p Þ; at the negation X k we substitute the negation

ðX 1 ; X 2 ; . . . ; X p Þ þ ðY 1 ; Y 2 ; . . . ; Y p Þ FðX k Þ ¼ X k G k and  is the XOR operation
¼ ðY 1 þ X 1 ; Y 2 þ X 2 ; . . . ; Y p þ X p Þ;
We remark that when Gk = 1 we have F(Xk) = Xk when
ðX 1 ; X 2 ; . . . ; X p Þ ¼ ðX 1 ; X 2 ; . . . ; X p Þ (12) Gk = 0 we have FðX k Þ ¼ X k For the negation F we have

That are the vector product, the vector sum and the vector FðX k Þ ¼ X k   
negation. In fact, FðFðXÞÞ ¼ FðXG ðXÞÞ ¼ ðXG ðXÞÞG ðXÞ ¼
For any vector we can compute the scalar product ðXG ðXÞ þ XG ðXÞÞG ðXÞ ¼ ½X þ G ðXÞ½X þ G ðXÞG ðXÞ þ
m1 X 1 þ m2 X 2 þ    þ m p X p ðXG ðXÞ þ XG ðXÞÞG ðXÞ ¼ XG ðXÞ þ XG ðXÞ ¼ X. So the
mðX 1 ; X 2 ; . . . ; X p Þ ¼ mðXÞ ¼ function F is an extension of the negation operation in the
m1 þ m2 þ    þ m p
classical logic.
¼ S1 X 1 þ S2 X 2 þ    þ S p X p We remark that in (15) we introduce a new logic variable Sk.
The vector G = (G1, G2, . . ., Gp,) is the inconsistent vector for
where m is the membership function and mk where k = 1, . . ., p
which the Tautology is not always true and the contradiction is
are the weights of any component in the Boolean vector. So the
not always false as in the Zadeh rule. In fact we have for the
membership function is the weighted average of the 1 or 0 value
contradiction expression
of the variables Xk.
We can show that C ¼ FðX k ÞX k ¼ ðX k G k ÞX k ¼ X k G k
when mðY 1 ; . . . ; Y p Þ  mðX 1 ; . . . ; X p Þ For the previous case the fuzzy contradiction C is not always
we have mðY 1 X 1 ; . . . ; Y p X p Þ ¼ min½mðXÞ; mðYÞ  mðY X̄Þ equal to zero. We remember that the classical contradiction is
X k X k ¼ 0 always. When G = (G1, G2, . . ., Gp) = (0, 0, . . ., 0) we
(13) came back to the classical negation and Ck = 0.
mðY 1 þ X 1 ; . . . ; Y p þ X p Þ ¼ max½mðXÞ; mðY 1 Þ þ mðY X̄Þ For the tautology we have
(14) T ¼ FðXÞ þ X ¼ ðX k G k Þ þ X k ¼ X k G k þ X k

Fig. 13. Three Boolean fields F1(A, B), F2(A, B), F3(A, B) for three agents.
 G. Resconi, M. Nikravesh / Applied Soft Computing 8 (2008) 1164–1177 1177

prototype fields. Morphic computing projects the external fields

to the internal context. In the projection process, the input is
reshaped in a way to select part of the input that is coherent with
the context. Because the projection operator uses the internal
prototype fields and the computed sources, we can give a
context dependent model of the external input. This is the
meaning of the new computation.

Fig. 14. Prototype agents.

References

The reference space is the space of all possible propositions [5] H.A. Fatmi, G. Resconi, A new computing principle, Il Nuovo Cimento,
and any agent is a Boolean field or prototype field. Given a vol. 101 B, No. 2, Febbraio 1988, pp. 239–242.
[6] D. Gabor, Holography 1948–1971, Proc. IEEE 60 (1972) 655–668.
Boolean field X in input, external agent, by the projection [9] J. McCarthy, M.L. Minsky, N. Rochester, C.E. Shannon, Proposal for
operator we generate the fuzzy field Y in output by the sources S Dartmouth summer research project on artificial intelligence, August 31,
as shown in Fig. 14. 1955.
Because in the projection operation we loss information the [12] R. Omnès, The Interpretation of Quantum Mechanics, Princeton Series in
Physics, 1994.
original boolean field for the input agent X is transformed in the
[13] Oshins, E., K.M. Ford, R.V. Rodriguez, F.D. Anger, A comparative
fuzzy field analysis: classical, fuzzy, and quantum logic, in: M.B. Fishman (Ed.),
QX ¼ Y ¼ mðXÞ ¼ S1 X 1 þ S2 X 2 þ    þ S p X p Presented at Second Florida Artificial Intelligence Research Symposium,
St. Petersburg, Florida, Advances in Artificial Intelligence Research, vol.
II, Greenwich, CT, JAI Press, Most Innovative Paper Award, FLAIRS-89,
Florida AI Research Symposium, April 5, 1989.
8. Conclusion
[16] G. Resconi, M. Nikravesh, Field theory and computing with words, FLINS
2006, in: Proceedings of the Seventh International FLINS Conference on
In this paper, we introduced a holistic interpretation of the Applied Artificial Intelligence, Italy, August 29–31, 2006.
computation denoted morphic computing. We show the [18] R. Sheldrake, A New Science of Life: The Hypothesis of Morphic
possibility of going beyond the traditional computation based Resonance (second edition, 1985), Park Street Press; Reprint edition
(March 1, 1995), 1981.
on a step-by-step process. Our type of computation is similar to
[19] R. Sheldrake, The Presence of the Past: Morphic Resonance and the Habits
a quantum measure where the instrument is the context with of Nature, Park Street Press, 1988, Reprint edition (March 1, 1995).
proper rules. Input fields are introduced in the instrument as a [23] P.Z. Wang, M. Sugeno, The factor fields and background structure for
context and write in internal parameters or sources of internal fuzzy subsets, Fuzzy Math. 2 (1982) 45–54.