1 / 27

Argyris Arnellos, Thomas Spyrou, John Darzentas University of the Aegean

ANALYSIS OF THE DESIGN OF A FRAMEWORK SUPPORTING MEANING PROCESSES IN LIVING AND ARTIFICIAL SYSTEMS. Argyris Arnellos, Thomas Spyrou, John Darzentas University of the Aegean Dept of Product & Systems Design Syros, Greece www.syros.aegean.gr. Background.

davidelopez
Download Presentation

Argyris Arnellos, Thomas Spyrou, John Darzentas University of the Aegean

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. ANALYSIS OF THE DESIGN OF A FRAMEWORK SUPPORTING MEANING PROCESSES IN LIVING AND ARTIFICIAL SYSTEMS Argyris Arnellos, Thomas Spyrou, John Darzentas University of the Aegean Dept of Product & Systems Design Syros, Greece www.syros.aegean.gr

  2. Background • Contemporary information systems research is focused on tackling the enormous quantity and complexity of information involved in these systems, usually resulting in the emergence of ill-defined problems. • We seek to analyse and reproduce human cognition in order to support the decision making. • Primary concentration on the cognitivist / connectionist, the dynamical and emergent paradigm, etiological frameworks. Note: Frameworks are differentiated by the way they handle the notions of intentionality, meaning, representation and information. One could ask: How is meaning generated and manipulated in natural and artificial cognitive systems.

  3. MAIN APPROACHES to GOGNITION and to the DESIGN of ARTIFICIAL AGENTS Cognitivism and Fodor’s Information Semantics • Representational content = that what a representation is supposed to represent. • Representation has some sort of informational correspondence to what is supposed to represent. The Problem: • A representation should be able to be in error. • Differentiations per se have no content. How can we account for the possibility of representational error when the informational correspondence does not exist and therefore cannot be correct? • The error should be detectable by the system itself. • Cognitive systems are not able to determine what their representations are supposed to represent.

  4. MAIN APPROACHES to GOGNITION and to the DESIGN of ARTIFICIAL AGENTS Computationalism: Cognitivism and Artificial Agents • Purely symbol systems with a predetermined and rule-based correspondence with state of affairs in the environment. • separate syntax and semantics • cognition in an algorithmic manner • All intentional content is a kind of information which is externally transmitted by a merely causal flow. • Meaning is externally given  system exhibits no autonomy and functionality other than that of its designer.

  5. Computationalism: Cognitivism and Artificial Agents • The level of “the fundamental syntactic entities” (symbol tokens) of an artificial system coincides to the level of its “fundamental semantic entities” (representations). • Searle’s Chinese Room Argument • Semantics are not intrinsic to syntax • Syntax is not intrinsic to physics (natural systems) • The symbol grounding problem is how to ground symbols in nothing else that meaningless symbols (Harnad, 1990). • Hybrid system where symbolic representations are grounded in non-symbolic one (iconic and categorical) • A lot of proposals in this direction, but with no success. Note: A complete cognitivist grounding theory should consider both external and internal representational content, as well as, their transduction system and its interactive nature. Main candidate for the transduction is a connectionist network.

  6. Etiological Frameworks of Cognition Millikan’s model of proper functions • A representation represents whatever it is its proper function to represent and it represents falsely if that content is false of a current representational target. The Problem: • A model of representation should not be epiphenomenal • Modeller's evaluation cannot count for a naturalistic explanation of the function of that representation. The satisfaction of her criteria are only causally justifiable. • The error should be detectable by the system itself. • Cognitive systems do not have access to their own relevant evolutionary history to be able to determine what the representations are supposed to represent.

  7. Etiological Frameworks of Cognition Dretske’s causal model An internal state “C” is recruited as a cause of a behavior “M” because of what it indicates about external conditions “F”, the conditions on which the success of M depends. Therefore, “C” becomes a representation of “F”. The Problem: • A model of representation should not be observer- dependent. • Dretske claims that “C” is a representation of “F” because this indicative relationship with “F” is a necessary part of a full explanation of why “C” has been recruited as “M”. • The error should be detectable by the system itself. • Cognitive systems do not have access to their own relevant evolutionary learning history. Dretske’s model cannot be defined in terms of current state.

  8. Dynamic Systems and Cognition • Time-Dependence: Natural cognition happens in real time, hence dynamics is better suited to model it than the a-temporal computational approach. • Embodiment: Cognition is embedded in a nervous system, in a body, and in an environment, whereas computationalism typically abstracts this embeddedness away, and can incorporate it in only an ad hoc manner. • Emergence: Dynamics can explain the emergence and stability of cognition through self-organisation, whereas cognitivism ignores the problem of cognitive emergence.

  9. Dynamic Systems and Cognition • No information processing (no symbols, no representations) • The dynamics of the cognitive substrate (matter) are taken to be the only thing responsible for its self-organization, and consequently for the system’s behavior. • System’s ability for classification is dependent on the richness of its attractors, which are used to represent events in its environment. • System’s meaning evolving threshold cannot transcend its attractor’s landscape complexity.

  10. Dynamic Systems and Representation • As tasks become more complex the use of internal states that carry information about the environment becomes less and less avoidable (Kirsh, 1991) • Brooks (1997): even in the very simple cases we find that individual units act as very simple representations in mediating interactions between the robot and its world • No representations per se, but a different type of representation, (Bichard,1998), (Beer, 2002)

  11. Interactivist Framework of Cognition • The grounds of cognition are adaptive far-from-equilibrium, recursively self-maintenant autonomy. • An appropriate infrastructure will exhibit both energetic and informational aspects. • concerned with the accomplishment of process switching and the accomplishment of the processes to which the system can switch. • Concerned with process switching. • Function emerges in such systems. • Representation emerges as a particular kind of function, as indications of potential interactions in the system. • Representation emerges in the evolution of higher levels of organisation. • Such a framework for cognition should be definable in terms of current state (including memory), not dependent on observer’s ascriptions or explanations. • Future-oriented, not past-oriented

  12. Interactivist + Constructivist Approach • No hybrid dynamical/cognitivist approaches. • Reconstructing cognitive concepts in a dynamical situated context. • This account conceptualises the‘aboutness’ of intentionality in terms of the complex relational structure of adaptiveinteraction processes. • The relation between the processes by which action is generated and theconstraints it must satisfy constitutes the basic way an organism orients itself in its worldand is thus the basic form of intentionality.

  13. Interactivist + Cosnrtuctivist Approach • Interaction: should shift away from the traditional focus on covariance between system states and environmentstates and include • downstream modulation of action and • interactive differentiation ofadaptive relations • Need of a deeper specification of self-organization processes

  14. 2nd order Cybernetics and Information • The complexity of an emergent self-organized system can only be enriched through its interaction with other systems in its environment. • Structural coupling increases the ability of the system’s classification. • But, second-order cybernetic systems admit no functional usefulness to representations and they regard information only as socially ascribed to a process from other observers.

  15. 2nd order Cybernetics and Information • What kind of representations? • representations that can only be understood in the context of activity. • Complex tasks may require active self-directed adaptive behavior • An asymmetry between system and environment requires intentional and goal-driven processes. Need to understand the relationship between intentionality, meaning, information and representations.

  16. PEIRCIAN SIGN PROCESSES • In a semiotic process a sign, or representamen, is something which stands to somebody for something in some respect or capacity. • It creates in the mind of a person an equivalent sign, or perhaps a more developed sign, which is called the interpretant of the first sign. • Each sign stands for its object, not in all respects, but in reference to a sort of idea, which is called the ground of the representamen. The sign-vehicle is the representative element, the foundation over and above which, a relation arises. In case of an absence of such a relation, the foundation is just a self-representation or object. sign-vehicle is the representamen

  17. Semiosis and Mediation (1 of 3) • For information carriers to be properly integrated and inform the representations of an adaptive system, the whole process of interaction must not be a mere sequential transmission, it must be a transformation. Two processes are needed: • one which producestextand one which can read it, while the reader-text relation must be of an active-mediative one. • a active mediation in which, the form of the input of the text is included and simultaneously expanded into the form of the output of the reader by means of the form of the output of the text. • Obviously not the case in the Chinese Room or in a Turing machine. • only substitution and not inclusion

  18. Semiosis and Mediation (2 of 3) • The dynamoid object would be the source of mediation so, it would determine the interpretant. • The determination is realized in a form and is the ground of the representamen. The ground should only be understood as form, for only as such can preserve the characteristics of the source, while allows it to be realized by a different process. • the mediator (representamen) will exhibit this form or ground by means of some qualities, the properties and relations it has independently of whether it serves as a mediator.

  19. Semiosis and Mediation (3 of 3) • Qualisign: the mediator shares certain qualities with what it mediates (the sign-vehicle) • Sinsign: mediator realizes at this very time the form of what it mediates • Legisign: mediator’s pattern realizes the form of what it mediates. The mediator’s ability for content by form inclusion gives the ability to expand the included form, which is to be interpreted by a cognitive system. On the other side of the relation (its relation to the interpretant) the mediator can be schematized in either of the three following types: • Semes:considered as the threshold of meaning, or the most primitive level of meaning, since they represent the point in the process at which the mediator can address an interpretant, or can cause the interpretant to interpret it (Rhemes) • Phemes: mediator has been formed in such a way that it can become information-giving for interpretants (Dicents). • Delomes: mediator has been formed in such a way as to become a provocative inference (Arguments)

  20. Semiosis and Intentionality As soon as the system represents the object based on information gathered by the representamen, the intentional action is completed. • In order for the system to be able to be intentionally related to its object then there must be a kind of interpretation for the mediator to be read as a representamen of its source, and not as the source itself. • Otherwise, representation cannot take place, or if it takes, it may be a mere causal process, a mere substitution as the one which appears in the systems of classic AI. • Then, the system will be able to build a representation of the object. The result of the representation gives the system the kind of the representation that the mediator supports for its source: • Icon: whatever the type of sign-relations of the mediator, they are taken to be similar to its dynamoid object. • Index: they are read as being temporally connected with the source • Symbol: there is a conventional and habitual connection with its object (organically or conventionally)

  21. Semiosis and Meaning • When the mediator has been schematized so as to become a rheme, the system will apprehend it as the immediate interpretant. • For the mediator to mean something for the system, the latter must be able to correlate the mediator’s schematization with the source (dynamoid object). • The sense that the system earns from the rise of II, would be temporally tested, so that it can go from Firstness to Secondness and become real meaning. At this point, the immediate interpretant become a dynamical interpretant and it is this moment that information about the source begins to have a semantic effect on the system. After the formation of the dynamical interpretant the self-organizing system has undergone a structural coupling with its environment.

  22. Meaning formation Rule-based syntactic complexity PS PS FI FI object action PS PS Morphodynamics sos of representational structure pragmatics system’s history DI Testing anticipations DO DO signs DO syntactics semantics II II IO DO measurement Formal rules ENVIRONMENT COGNITIVE SYSTEM

  23. Grounded Interactions in Pragmatic Meaning • Interaction initiates from the (DO), the environmental element of interaction, which needs mediation in order to be determined in a representation of the cognitive system. • A direct relation between a DO and its interpretant is impossible • Use of a means of measurement, the nature of the DO can be indicated by a primitive internal representation, the immediate object (IO). • the sign (via the sign-vehicle), indicates the direction of the “reality” to which it refers. • A sign-vehicle contains several IOs which in turn refer to several DOs. Which IO will eventually be actualized depends on the cognitive system’s anticipations.

  24. Grounded Interactions in Pragmatic Meaning • Semantics: They make use of the given signal in order to infer its meaning, but only in the sense of the ground of the representation. • As the form of the ground is a function of the qualities, properties and relations of the sign-vehicle, the latter constraints the former, which sets the borders for the structure of the impending representation Narrowing the semantic content • the result is tested against the system’s anticipations • an interpretation of the parameters of the sign-vehicle in a way narrow the IO’s selection and give a certain directionality to the system. The immediate interpretant (II) has been formed.

  25. Grounded Interactions in Pragmatic Meaning • For II’s formation system’s semantics must be tested against the pragmatics for the ground to be enriched with -not all possible- but only system’s relevant and useful (at the specific moment) predicates. • There will be many tests needed in order this core meaning to be temporarily stabilized into a dynamic interpretant (DI). • This process is the most complicated one as it requires the properties of the self-organizing part of the system Requires self-reference and functional closure since the system must refer to itself in order for unsuccessful structure modifications to be obliterated

  26. Grounded Interactions in Pragmatic Meaning • Meaning of a representation is given by the system’s expectations involving its object. • Content of a representational structure could be said to be the intersection of the information sets of all these expectations and • this content provides the information needed to reason with the respective representation • Meaning, which results from the semantic processes, should be open to revision. • Morphodynamical processes to dynamically manipulate meaning structures in terms and by means of forms, which are simultaneously tested against the system’s anticipations (pragmatics) within the conditions of the functional closure offered by the dynamics of the system.

  27. Grounded Interactions in Pragmatic Meaning Pragmatic Meaning-based interaction: • When the system has been informed by its interaction, in such a way, that its new dynamic state is compatible with the system’s representations, then a meaningful interaction has been achieved. • The compatibility is immediately connected to the fitting of the new form with the system’s functions, representations and structural constraints. • Under this view a self-organizing system uses information from the environment (or of its own). • The new forms will self-organize reaching certain levels of discontinuity, while newer ones will come and carry on to new organizations.

More Related