poniedziałek, 7 października 2013

Artificial Intelligence University

News: I'm converting this into a "page", since a "post" is more of a finished thing genre. New updates will only go to the page version: http://lukstafi.blogspot.com/p/ai-university.html

Most of the links come from Video Lectures. The thesis is that the courses available online can form a solid education in AI. I have updated the list to provide a more balanced program, aiming at "university replacement". Tentatively one could go through four courses in a semester. I will add links to textbooks later.

  1. year:
    1. Introduction to Logic by Michael Genesereth
      1. General Game Playing by Michael Genesereth, also Michael Thielscher and Sam Schreiber -- at this point, take a quick tour by watching the (short and simple) lectures
    2. Probability Primer by mathematicalmonk
    3. Calculus: Single Variable by Robert Ghrist
    4. Algorithms, either one of:
      1. Algorithms: Design and Analysis, Part 2 by Tim Roughgarden
      2. Algorithms, Part 2 by Kevin Wayne and Robert Sedgewick
      3. Introduction to Algorithms by Charles Leiserson and Erik Demaine
    5. Functional Programming Principles in Scala by Martin Odersky
    6. Machine Learning  by Andrew Ng
    7. Introduction to Cognitive Architectures seminar:
      1. Cognitive Architectures by Włodek Duch
      2. Clarion Tutorial, Clarion Part 2 by Michael Lynch
      3. The Soar Cognitive Architecture  by Nate Debrinsky
      4. OpenCog by Ben Goertzel
      5. From Constructionist to Constructivist A.I. by Kristinn R. Thórisson
      6. Deconstructing Reinforcement Learning in Sigma, Modeling Two-Player Games in the Sigma Graphical Cognitive Architecture by Paul Rosenbloom
      7. Pursuing Artificial General Intelligence By Leveraging the Knowledge Capabilities Of ACT-R by Alesandro Oltramari
      8. A Cognitive Architecture based on Dual Process Theory (perception vs. imagination) by Claes Strannegård
    8. Scientific Approaches to Consciousness  by John F. Kihlstrom
  2. year:
    1. Probabilistic Graphical Models by Daphne Koller
    2. Course on Information Theory, Pattern Recognition, and Neural Networks by David MacKay
    3. Introduction to Modal Logic by Rajeev P. Goré
    4. Introduction to Databases  by Jennifer Widom
    5. Learning From Data  by Yaser Abu-Mostafa (Machine Learning with elements of Statistical Learning Theory)
    6. Linear Algebra by Gilbert Strang, also:
      1. Complex Analysis by Petra Bonfert-Taylor (optional)
      2. Differential Equations by Arthur Mattuck (optional)
      3. Introduction to Functional Analysis  by Richard Melrose (optional)
      4. Nonlinear Dynamics I: Chaos by Daniel Rothman (optional)
      5. Differential Geometry  by Paul Seidel (optional)
      • The optional math classes are meant to be picked up later as your time allows. You should at least have basic familiarity with: complex numbers; calculus and differential equations; linear operators: matrix representation in various bases, nullspaces, orthogonal complement.
    7. For a round number of courses, pick one more of the math courses above
    8. Introduction to Philosophy  by Richard Brown
  3. year:
    1. Discrete Optimization by Pascal Van Hentenryck
    2. Artificial Intelligence Planning by Gerhard Wickler and Austin Tate
    3. Introduction to Formal Languages, Automata and Computational Complexity  by Jeff Ullman
    4. Natural Language Processing, one of, or both:
      1. Dan Jurafsky and Christopher Manning
      2. Michael Collins
    5. Neural Networks by Geoffrey Hinton
      1. and Neural Networks class by Hugo Larochelle
    6.  Either:
      1. Machine Learning (review and continuation) by Andrew Ng, or
      2. Introduction to Machine Learning by Alex Smola.
      • Skip over parts that you are confident to know already.
    7. Linear Dynamical Systems by Stephen Boyd
    8. Computational Neuroscience by Rajesh P. N. Rao and Adrienne Fairhall
  4. year:
    1. Game Theory  by Kevin Leyton-Brown, Matthew O. Jackson and Yoav Shoham (optional)
    2. General Game Playing by Michael Genesereth, also Michael Thielscher and Sam Schreiber -- at this point, treat it as a project course, build your own player using knowledge from other courses
    3. Convex Optimization by Stephen Boyd (optional)
    4. Reinforcement Learning -- sorry for redundancy with each other and with pieces in Andrew Ng, try to find your way
      1. by Csaba Szepesvári,
      2. by Satinder Singh Baveja,
      1. Foundations of Machine Learning by Marcus Hutter,
      2. Richard Sutton AGI 2010 Keynote Address, Part 2 
      3. GQ(lambda)- A General Gradient Algorithm for Temporal-Difference Prediction Learning with Eligibility Traces by Hamid Reza Maei
    5. Abstract Algebra  by Benedict Gross
    6. Overview of Automated Reasoning  by Peter Baumgartner
    7. Type Theory Foundations and Proof Theory Foundations by Robert Harper and Frank Pfenning respectively
    8. Ethics and Moral Issues  by Richard Brown
  5. year:
    1. Big Data, Large Scale Machine Learning  by John Langford and Yann LeCun
    2. Graphical Models and Variational Methods by Christopher Bishop
    3. Statistical Learning Theory by John Shawe-Taylor and by Olivier Bousquet / newer variant of Olivier's
    4. Practical Statistical Relational Learning by Pedro Domingos
    5. Online Learning, Regret Minimization, and Game Theory by Avrim Blum
    6. Introduction to Category Theory by error792
    7. Computer Vision by Mubarak Shah
    8. Cognitive Architectures and Modeling Course -- perhaps some combination of these, but there is no good course online:
      1. Representations: Classes, Trajectories, Transitions and Architectures: GPS, SOAR, Subsumption, Society of Mind by Patrick H. Winston, as introduction
      2. The Society of Mind by Marvin Minsky
      3. Cognitive Science and Machine Learning Summer School videos
      4. Cognitive Modeling by John Anderson and T.A. Phil Pavlik
        1. Cognitive Modelling by Sharon Goldwater
      5. AGI 2011 ArchitecturesPart 2 and other AGI Conference presentation videos

czwartek, 8 sierpnia 2013

Reading Gary Drescher "Made Up Minds"

The "schema mechanism" system developed in Gary Drescher's PhD project reminds me of Anticipation-Based Learning Classifier Systems, but it is more AGI / cognitive-architecture worthy because of its representation building facilities. The book is very novel for its year 1991, before modern RL theory became popular in AI. ABLCSes are quite recent development in LCSes (although with an early publication in 1990). I have (good) sentiment towards Learning Classifier Systems, they were my first encounter with a more cognitive form of AI, very long ago.

It would be cool to redo Gary Drescher's project, but using Rich Sutton and Hamid Maei's recent results -- Gradient Temporal-Difference Algorithms with off-policy learning -- instead of schemas.

The principle behind Gary's project is constructivism, the opposite of nativism. Almost all structure of the world, even what Kant claimed to be necessarily needed a priori -- binding of experience into objects having features -- can be learned from input by a relatively simple mechanism. You might think that therefore constructivism isn't only radically against Chomsky, but also David Deutsch'es ideas I've been quoting recently --  bashing of empiricism and of logico-positivism. But consider this: the construction algorithm, to succeed, must have universal reach in David Deutsch'es terms. And Gary Drescher accepts the criticisms against logico-empiricism, "Even the most rudimentary conceptions of the physical object cannot be defined by the schema mechanism as any function of the sensory primitives." Section 8.6 stresses "Why Non-naive Induction Must Be Built In"; and the mechanism needs to solve similar problems with counterfactuals. The system uses counterfactuals for learning and concept invention. "The difficulty and obscurity of the concept of counterfactuals is, I suspect, a reason that its fundamental importance for learning systems has been late to be recognized, rather than a reason to consider it an implausible basis for learning."

Notes below focus on chapters 3 and 4 as these describe the mechanisms of the cognitive system. I pick a couple of nuggets from later chapters. Chapter 2 presents Piagetian theory, elaborating on initial stages of development. It is worth reading in full.
  1. Conditions are conjunctions of items. Primitive items are sensory facts and synthetic items are beliefs.
  2. Primitive items are binary: On / Off, synthetic items are ternary: On / Off / Unknown.
  3. Objects and relations between objects are supposed to be stable configurations of schemas, synthetic items and composite actions.
  4. Schemas are identified by a triple: preconditions, action, postconditions.
  5. Primitive actions change state of the world and agent.
  6. Composite actions are identified by a postcondition they achieve.
  7. Therefore a schema might be a refinement of the postcondition of a composite action under given preconditions.
  8. Accessible value in a state is the maximum value achievable along a reliable path from the state.
  9. Instrumental value is assigned to items along reliable paths to a goal. I.e. items in a state have its accessible value as instrumental value. Instrumental value is transient.
  10. Delegated value is assigned proportionally to (1) the difference of average accessible values of an item: when the item is On minus when the item is Off; and (2) the duration of the item being On. It is more permanent.
  11. Item of frequently instrumental value only has delegated value when it is not "readily accessible" -- readily accessible items have the difference in accessible values (between their On/Off states) close to zero.
  12. To avoid runaway of propagating delegated value through cycles, the value propagated is half of the delegated value.
  13. Attention via hysteresis-habituation loop: recently selected schemas have more weight but decreasing with consecutive selections.
  14. Reweighting to upsample schemas with rare actions.
  15. Promote actions with inverse effects (turns an item On/Off right after it was turned Off/On) as it heuristically leads to reliable schemas.
  16. Marginal attribution:
    1. Start from bare schema: {}-A->{}; add results in spinoff schemas for items whose positive-transition (from Off to On) or negative-transition rates following action A are higher than their averages for all actions.
    2. Add (negated) items in contexts of spinoff schemas which (anti)correlate with validity of the schema, making the spinoff schema more reliable.
    3. All statistics count only unexplained transitions and correlations -- when the items aren't in results or contexts of existing valid schemas. (Statistics are reset after a spinoff accordingly.) Statistics are only collected by most specific schemas accounting for a situation. This increases sensitivity to regularities and reduces combinatorial explosion.
  17. Schema chaining requires that all items of the (primary) context of a following schema are provided by results of preceding schema. It's used for composite actions, etc.
  18. Schemas have associated extended context (and extended results). Extended contexts and results are mutable (evolve over time). Besides being a data structure for spinoff formation, extended context adds to the condition for activation of a schema. Schema chaining cannot rely on the corresponding spinoff schemas, because it often requires general primary contexts.
  19. For a composite action, besides conditions that need to hold initially, there can also be conditions that need to hold throughout the action.
  20. Synthetic items are designed to identify invariants when all apparent manifestations change or cease (compare Piagetian conservation phenomena).
  21. Keep track of local consistency: the probability that a schema will be valid right after it has been valid, and expected duration of consistency: how long since onset of validity to first invalid activation. (Recall attention via hysteresis.)
  22. A synthetic item is a reifier of validity conditions of an unreliable locally consistent schema, called its host schema -- action called probing action and result called manifestation. I.e. it is the state item such that when added to the context (precondition) of the schema, if the schema were activated, the action would bring the result.
    1. The intention is that the synthetic item captures a persistent feature, like presence of an object, while the remaining context items of the host schema capture transient features, like effector configuration.
  23. Learned verification conditions set the state of a synthetic item:
    1. Host schema trial: when host schema is activated, On resp. Of if it succeded resp. failed.
    2. Local consistency: the state remains as changed for at most a period of expected duration of host schema's local consistency (for On, local "inconsistency" for Off). Then revert to Unknown.
    3. Augmented context conditions: the extended context of the host schema (which collects evidence from spinoff schemas).
    4. Predictions: "If a synthetic item appears in the result of a reliable schema, and that schema is activated, then in the absence of any evidence to the contrary, the mechanism presumes that that schema succeeded".
  24. Above mechanism approximates a synthetic item, but the synthetic item is not coextensive with any function of cumulative inputs (i.e. of the inputs history).
    1. "The schema mechanism grounds its synthetic items in the reification of counter-factual assertions; the subsequent adaptation of its verification conditions is driven by that grounding."
  25. Composite action is created for each spinoff schema that has a novel result.
  26. "A composite action is considered to have been implicitly taken whenever its goal state becomes satisfied [...] Marginal attribution can thereby detect results caused by the goal state, even if the goal state obtains due to external events." Which together with hysteresis leads to imitation.
  27. Backward Broadcast mechanism and action controller learn proximity of schemas (results) to goal states (of composite actions). If reliable chains of schemas are found, they are incorporated into composite actions. The chains are also used for forward prediction.
  28. Action controller handles special cases: indeterministic actions (schemas with same contexts and actions but various results), repetition, on-the-fly repair (detecting schema that make applicable some component of an interrupted action).
  29. Schema with composite action cannot spin off a schema with part of the composite action goal in the results.
  30. There is no "problem resolution" mechanism. Rather, some schemas hit a dead-end, and are taken over by schemas that capture more fruitful regularities.
  31. I haven't understood how inversely indexed representations (synthetic items) work. (par. 6.4.4)
  32. Note that synthetic items do not represent identity of for example tactile-visual objects. This isn't bad though because the system's mistakes reproduce Piagetian errors at corresponding developmental stages. Errors mean unreliable schemas which leads to further development.
  33. Now on to more far-fetched stuff. "The new conception [learned abstract concept] reifies the set of circumstances under which a piece of one's computational machinery behaves a certain way."
  34. "Consciousness requires knowledge (and hence representation) of one's own mental experiences as such; the schema mechanism does not come close to demonstrating such knowledge."
  35. Unimplemented mechanism: subactivation. "To subactivate an applicable schema is essentially to simulate taking its action, by forcing its result items into a simulated-On state (or, if negated, a simulated-Off state)." The simulated states are entirely distinct from actual states and all mechanisms are duplicated for them. But statistics are shared, and spinoff schemas are created "for real".
    1. Simulations are serial but parallel chaining search will cache the knowledge.
  36. Unimplemented mechanism: explicitly represent inverse actions to make them available for subactivation (i.e. simulation).
  37. Override generalizations: when a derived (by simulation) schema prediction is wrong because a direct schema from which it was derived is overriden, derived schema should be overriden too, without penalty for the derivation. A new schema will be created to capture this exception.
    1. "The suggestion is that deductive-override machinery may permit the schema mechanism to escape the fallacy of naive induction. The key is to regard the conflict between a reasonable generalization and an absurd but always-confirmed generalization as just another conflict between generalizations expressed at different levels of description."
  38. "The reason [not to build in] a variable-matching implementation of generalizations, is just that there is no apparent way to support such an implementation without abandoning the constructivist working hypothesis by including domain-specific build-in structure. [...] Perhaps the system itself could be designed to devise explicit structured representations to support variablized generalizations. [I]f virtual generalization fails, devising such machinery may be vital to the schema mechanism."
  39. "A schema's extended context is essentially a connectionist network solving a classifier problem."
  40. Unimplemented: clustering (i.e. hierarchical modeling); "having coordinated coarse- and fine-grained spaces mitigates the combinatorics of showing the path from one fine-grained place to another, because the path can be represented as a coarse segment to get in the right vicinity, followed by a fine-tuning segment."
  41. Unimplemented: garbage collection. Schemas: not contributing to goal achievement, not spawning new spinoffs, seldom activated, perhaps even those that are activated but recreation opportunities are more frequent than activation opportunities.

piątek, 22 lutego 2013

Dump of my recent "The Hard Problem" comments


  • Ascription of consciousness is a matter of degree, i.e. an entity can be slightly conscious or very conscious, with a lower bound of no consciousness but probably without upper bound. Consciousness is a complex process, a complex of processes of consciousness. For the most part, processes of consciousness are about representation and action (some like to call it intentionality). A necessary condition for a given degree of consciousness is a given degree of complexity of adequate representation including representation of actions (i.e. bidirectional causality between processes of consciousness and remaining processes of environment). Consciousness is relative to the environment of which it is conscious, i.e. you judge reality of consciousness by how you judge the reality of its environment (e.g. parallel universe, formally stated complete description of a universe, ...) Sufficient conditions for a given degree of consciousness are satisfaction to a given degree of constraints of the kind described in "Being No One" by Thomas Metzinger. Consciousness is not fundamental to all life, it does not necessitate biological life, and it is not epiphenomenal. Current computers don't have mental processes (AFAICT).
  • I mean by it that single cellular organisms do not have consciousness, and an "epsilon" of consciousness only appears in animals with a central nervous system. On the other hand, consciousness can for example be running on a set of integrated silicon-based chips created by humans, and be sent to a planet composed of only inorganic materials.
  • The notion of potentiality is important for this view. Mental processes are processes that are potentially conscious. Dreams and mental imagery are conscious in so far as they are potentially representative (or pertain to such). Also, I don't mind using the word "presentation" rather than "representation" in "transparent" cases -- where we don't have knowledge of our experiences, but rather have knowledge of the experienced things by having the experiences.
  • [Epiphenomenalism] is logically impossible (once you define consciousness somehow), but it is conceivable. Imagine that you stare on a webcam attached to someone else's head. Now imagine that a brick is falling on that person. You instinctively duck the brick. As it turns out, that person ducked the brick in the same manner. Suddenly you feel you are the person who sees the world through the webcam (due to the accidental correlation of your intention with the behavior). Epiphenomenalism would be very similar to this scenario, only with a systematic explanation of the "accidental correlation" via the claim that intention is just a manner of perceiving the onset of an action.
  • In this analogy there is the causally active consciousness of the person with the webcam, and the epiphenomenal consciousness of the person watching, insofar as the webcam transmission is concerned. [...] a prototype/analogy for a single experience where the causal link from the experience to its physical carrier is broken. To build a full-fledged epiphenomenalism you need to limit all experiences to such a kind.
  • Consciousness can the primary reality of the universe in diametrically different ways. (1a) It can be entirely different substance from material stuff -- the abandoned dualism. (1b) It can be an aggregation of an aspect of physical stuff that has "propensity for subjectivity", for example "quantum collapsability" or some other spookiness. (1c) It can just be some particular biological material processes but not others (the sense of "primary reality of consciousness" here is that we cannot "explain it away" structurally/functionally, but could at best duplicate these processes).
  • (2a) Epiphenomenalism is just a very narrow case of consciousness arising from material processes, one where it does not have causal effects (it only has causes). There are at least two other possibilities: (2b) consciousness is a particular coarse-graining of material processes, the effects of those processes are the effects of consciousness (with emphasis on the structure of the coarse-graining, not on the particular processes as in 1c); (2c) consciousness is a particular functional structure of material processes among other processes, the functional structure exhibiting so called downward causation on the processes that exhibit it.
  • "The neurological explanation of all measurable events related to 'orange' involves numbers and theories and chemistry, no need for the perception known as 'orange'." Well put. This is also known as "the problem of qualia", although I prefer calling it "the hard problem" because the notion of qualia is not really that clear-cut. For example "Normal listeners can discriminate about 1,400 steps of pitch difference across the audible frequency range, but they can recognize these steps as examples of only about 80 different pitches." I can discriminate two shades of orange when I see both nearby, without being able to give them names because when separated in time they just seem the same to me.
  • Perhaps a good way to start people thinking about this is to go back to B. Russell "Analysis of Mind" (1921). First, realize that when you note that you see an orange patch, you do not learn that you have an orange experience, you learn that you have experience of orange. "If there is a subject, it can have a relation to the patch of colour, namely, the sort of relation which we might call awareness. In that case the sensation, as a mental event, will consist of awareness of the colour, while the colour itself will remain wholly physical, and may be called the sense-datum, to distinguish it from the sensation. The subject, however, appears to be a logical fiction, like mathematical points and instants. It is introduced, not because observation reveals it, but because it is linguistically convenient and apparently demanded by grammar. [...] But when we [dispense with the subject, we cannot distinguish the sensation from the quality of the property.] Accordingly the sensation that we have when we see a patch of colour simply is that patch of colour, an actual constituent of the physical world, and part of what physics is concerned with [...]" In "My Philosophical Development" where I quote from, Russell continues, "It became possible to think that what the physiologist regards as matter in the brain is actually composed of thoughts and feelings, and that the difference between mind and matter is merely one of arrangement." This is close to the view that John Searle has. (My allegiance is more with Searle's opponents though, with representational functionalism of for example Daniel Dennett.)
  • Some philosophers are obsessed with the idea of “intrinsic existence”. We say that consciousness is (where “is” and “arises” have the same meaning) the structure of complex processes of representation and action. They say, structure is a relation among elements, and structure/relations do not have intrinsic existence because they are what is preserved between the existing stuff and arbitrary representations of the organization of the stuff, given an instruction of interpretation. Therefore they see our claim that consciousness is a particular structure of processes as equating something that has independent existence, with something that is a result of analysis/interpretation.
  • What we mean is functional structure, that consciousness is the structure of causal interaction within a system and across the system and the environment. They ridicule the idea that consciousness floats out of the brain sustained by causal links with pieces of environment. They say that besides, it doesn’t save function from being “in the eye of the functionalist”, that functions are ascribed rather than independently existing.

piątek, 7 września 2012

Reading Thomas Metzinger -- metaphysics


Metzinger's metaphysical argument against the existence of mental
notions such as the self is weak. It is mostly a proof by assumption
of the thesis, as Metzinger starts on materialistic
grounds. Metzinger's position is roughly that of eliminativist
materialism, as in Richard Rorty's "Philosophy and the Mirror of
Nature" (chapter 2) I'm just reading. Metzinger seems to employ the
developed theory to provide what Rorty calls "criteria for the
identity of reference". Rorty moves past eliminativist materialism:
"But since I think that the reductive and eliminative versions of
the identity theory are both merely awkward attempts to throw into
current philosophical jargon our natural reaction to an encounter
with the Antipodeans, I do not think that the difference between the
two should be pressed. Rather, they should both be abandoned, and with
them the notion of "mind-body identity." The proper reaction to the
Antipodean story is to adopt a materialism which is not an identity
theory in any sense, and which thus avoids the artificial notion that
we must wait upon "an adequate theory of meaning (or reference)"
before deciding issues in the philosophy of mind."

To see why Metzinger's eliminativist argument is weak, it is helpful
to consider the effect of Metzinger's "identity criteria" on the
Antipodeans scenario. Antipodeans are people just like humans, only
they do not use mental terms at all. At the point of cultural
development when they started to feel the need to talk about their
thoughts and perceptions rather than directly about things in the
world, their biological sciences were already developed well enough
that they could employ terms referring to brain states. Having
appropriate medical instruments, they taught themselves to name their
states such as "pain" by what they are, e.g. "activations of
C-fibers". For Rorty, the scenario is fully plausible, and in the
scenario there is really no difference between Antipodeans and humans
outside of their "folk psychology". For Metzinger, the scenario is
implausible, for the Antipodean terminology is indicative that their
phenomenal life is much different than in the human case, Antipodeans
probably are system-conscious. So large shift in their phenomenal
experience would project on the rest of their culture.

The point of eliminativism in Richar Rorty's words is: "Some
statements of the form "I just had a sensation of pain" are as
properly taken as true as "The sky is overcast" and "The sun is
rising," but none of them is true." The identity of reference criteria
that Metzinger sets out to provide are like the theory explaining how
"The sky is a blue dome" is false, by explaining what the sentece
speaks about, and how the facts of the matter differ from facts the
sentence postulates. But once Metzinger is done building the referents
of "pain", instead of saying "and here is how the talk about
sensations and selves gets things wrong", he claims "and since we can
now see that sensations and selves are complex processes rather than
simple substances, they do not exist" -- wait, what?

Reading Thomas Metzinger -- intentionality

Based on Thomas Metzinger "Being No One", starting from section
6.4.3. What is intentionality anyway?

"[T]he content of a perceptual state really is not a part of the
environment, but a relation holding to this part [...] Full-blown,
phenomenal self-consciousness always involves a relation between the
self and an object component."

"Some stages [of attentional agency] are conscious, some are
unconscious.  As a whole, this process displays an extremely high
degree of flexibility and short-term adaptability, involving the
explicit internal simulation of alternative objects for attentional
processing. We like to call this “selectivity,” [...] What there
is, in the sort of phenomenal agency involved in focal attention, is a
globally available representation of the process in which different
alternatives are matched against each other and the system settles on
a single solution."

"If a system integrates its own operations with opaque mental
representations, that is, with mental simulations of propositional
structures that could be true or false, into its already existing
transparent self-model while simultaneously attributing the causal
role of generating these representational states to itself," wait, how
does one do that?

"Opaque mental representations" are simply those that are not perceived as standing for reality. There is conscious meta-representation, but it doesn't have dedicated channels. It is just the knowledge that some experiences are not veridical -- for example the doubling of the world when one presses the sides of the eyeballs. And it is also the knowledge of how to use mental faculties to manipulate higher, more abstract layers of "modality stacks". Opaque are the representations that arise by top-down modulation instead of being driven by the inputs (the senses, the motor feedback). "[In volitional thought] the object component is opaque. We know that we take a certain attitude toward a self-generated representation of a goal."

"Please note how a phenomenal first-person perspective now reveals itself as the ongoing conscious representation of dynamic subject-object relations: to see an object, to feel a pain, to selectively “make a thought your own,” to chose a potential action goal, or, to be certain of oneself, as currently existing." Metzinger leads me to the conclusion that we have a self-perception modality, as I mentioned at the end of foregoing comment. The subject-object relation character of experience comes from cross-modal binding with this modality. But... "Cognitive self-reference, therefore, on the phenomenal level is necessarily experienced as direct and immediate, because it is not mediated through any sensory channel (it takes place in a supramodal format) and because of the fact that it is a second-order process of phenomenal representation, is not introspectively available (naive realism)." So is that a wrong conclusion, is self-perception supramodal? Although he speaks here about the cognitive layer, the "channel" refers to the earlier (phenomenal) layer of what I might call "modality stack". The supramodal aspect is just the more abstract entities arising from cross-modality binding in higher layers. Modalities correspond to senses, so perhaps we need a different term because we also need to cover "effector stacks", the layers of the "motor cortex".

"In short, phenomenal models of the intentionality relation consist of a transparent subject component and varying object components, which can be transparent as well as opaque, transiently being integrated into an overarching, comprehensive representation of the system as standing in a specific relation to a certain part of the world. [...] Episodic memory is a process of reconstructing what was here termed a PMIR, because one necessary constituent of memory retrieval is not simply the simulation of a past event, but an association of this simulation with a self-representation. [...] Reactivating a PMIR inevitably means reactivating a PSM." Patients without the PMIR are zombie-like. "Akinetic mutism is a state of wakefulness, combined with the absence of speech, emotional expression, and movement. Obviously, for such patients there is an integrated functional self-model, because they are able to briefly track objects, pull a bedcover, or, if forced, say their own name. [...] What the outside observer experiences as a vacuous stare or emotional neutrality is the complete absence of any willed action or communicative intention, the absence of a globally available model of subject-object relations (and, as can be seen in lacking the desire to talk, of subject-subject relations as well)." 

"The experience of agency seems to be the ongoing representational dynamics collapsing a phenomenal model of the practical intentionality relationship into a new transparent self-model. [...] It is important to note that at least two different kinds of ownership will be involved in any accurate description of the phenomenology: ownership for the resulting body movements [PSM], and ownership for the corresponding volitional act [PMIR], for example, the conscious representation of the selection process preceding the actual behavior." The PMIR is an integrative capacity.

I haven't read chapter 7 but PMIR seems to me to just describe various binding processes where one of the components is the transparent part of the PSM. I don't think these binding processes form a very distinctive separate module. If you describe the structure of the PSM, with transparent lower layer and more abstract higher layers etc., and describe the binding processes in general, there doesn't seem much to add specific to the PMIR. ETA: binding is not the right term since I meant processes that integrate across objects, binding refers to integrating experiences into objects. This large class of integrative processes, complementary to binding processes, deserve a name. Might it be PMIR? But these processes are all over the place, by themselves they don't model anything. And the integrative processes were already covered in the first part of the book. Perhaps Metzinger's thesis is that there is something special about this subclass of them.

piątek, 17 sierpnia 2012

Reading Thomas Metzinger -- the self


What is the role of the self? Based on chapter 5, (mostly)
subchapters 6.1--6.3, and section 6.4.2 of "Being No One". Note that I
only cover stuff where I want to add (or question) something, many
interesting things get omitted.

Phenomenal Self Model


It seems that the Phenomenal Model of the Intentionality Relation is a
structure and process for managing attention. When an object is
integrated into PMIR, the directing of attention towards that object
is available phenomenally (i.e. available for action control, concept
formation and "higher order" attention). PMIR will be basis of next post.

Integrating the Phenomenal Self Model into the world model provides
for relations between the PSM concept and the concepts of "outside"
objects. Such relations are basis of goal-directed behavior (when a
simulated relation is different than an actual relation).

The PSM integrates also the representation of any process that
preattentively integrates a set of features as an object. Only then
the object becomes attentionally available, so this is even more basic
than PMIR, it enables all "higher order" relations. I'm not sure
whether Metzinger thinks the object is available for concept formation
prior to this integration into PSM, it is unlikely given the lack of
attention. But for sure, what is here integrated is the distinct
concept of perceiving the object. Now it becomes possible to have the
goal of "taking a better look" at an object. This is a considerable
limitation over how I imagined attentional availability, but since a
global availability means integration into world model, it shouldn't
be much harder to also integrate it subjectively, i.e. into the
PSM. The very object formation already happens in the world-model and
the PSM, and in this process the object-encoding processes become
mental representations, i.e. potentially globally available.

PSM is distinguished in the representational space by high
invariability. Attention (for PSM and generally) might be driven to
"places" where the output of the model and the input of the senses
differs the most. But the background self-awareness is always present.

All properties of the system are represented in one integrated data
format, i.e. the PSM is "holistic".

"The nervous system, and this will be true for the particular case
of self-representation as well, is not so much a top-down controller,
but more a system whose task consists in generating adequate patterns
within [the body - nervous system - environment] overall
dynamics. [...] [I]f an internal self-model is to be a successful
instrument in predicting the future behavioral dynamics of an
individual system, it must in some way mirror or reflect important
functional characteristics of precisely that part of its internal
dynamics, which in the end will cause the overt actions in question."

"A fully grounded self-model would simply disappear. In principle,
phenomenal selfhood emerges as long as there is a conflict or
incoherence between bottom-up and top-down processes, between
expectancy and actual perception." The discrepancy draws attention to
the respective emulators in general, making them conscious,
i.e. globally available for flexible reaction. "A certain level of
autonomous, residual self-modeling is preserved."  Note that dynamics
(including goal-directedness) is part of the self-model.

Revisiting Transparency


Metzinger makes a general comment that I might have missed in my
discussion of transparency (don't remember). Content is phenomenally
opaque when it is presented as representational, i.e. as correlated
with a different presented content (for example, an imaginated
rehearsing self is correlated with the actual self integrated into the
"Now"). But later in the section he goes on with the old confusion :-(
First, by having a "fully opaque" self experience being an experience
of a ghost or spirit (disembodied entity). Later, he seems to say that
opaqueness is the "presentation" of misrepresentation in
presentational content. But then he writes "There is always
self-presentational content, there are always emotions and gut
feelings, for instance, and presentational content is always fully
transparent." Yet this is specifically about the self, so that's
OK. It looks like "presented as misrepresentational" is achieved by
having the relation, but missing the target process with which the
opaque content is supposed to correlate.

This transparency confusion makes it difficult to decipher notions
such as "nemocentric reality model (centered on a globally available,
but fully opaque self-model embedded in the current virtual window of
presence)", because here it certainly is not about presenting
misrepresentation (in neither the simulation nor the
pseudo-hallucination form). It is also not (or is only to a degree)
about representing an independently presented content (neither in the
simulation sense nor in the "constitutionally earlier stages of
processing" sense). Although I have an intuition what "fully opaque"
is supposed to mean here, where is the meat? "For any phenomenal
representation, its degree of phenomenal opacity is given by the
degree of attentional availability of earlier processing stages."
Yeah, you've already told that... OK, I uderstand "fully opaque"
as that the dependency of every processing stage on an earlier stage
(or generally -- bidirectionally, on other stages) is represented.

"My hypothesis is that the phenomenon of transparent self-modeling
developed as an evolutionary viable strategy because it constituted a
reliable way of making system-related information available without
entangling the system in endless internal loops of higher-order
self-modeling." No, the solution was already flourishing before the
problem developed :-)

Transparency finale: "A transparent representation is characterized by
the fact that the only properties accessible to introspective
attention are their content properties. It does not allow for the
representation of a vehicle-content distinction using on-board
resources. [...] If I engage in typical cognitive activities like
reasoning, and if I then direct my introspective attention to this
process as it unfolds, I experience myself as operating with internal
representations that I am deliberately constructing myself. They do
not imply the existence of their simulanda, and they might be
coreferential with other mental representations of myself without me
knowing this very fact. [...] [T]he phenomenology of transparent
experience is the phenomenology of not only knowing but of also
knowing that you know while you know; opaque experience is the
experience of knowing while also (nonconceptually, attentionally)
knowing that you may be wrong."

Role of Objects


"Global availability of information always means availability for
transient, dynamical integration with the currently active
self-model." I don't know what's the added value of PMIR over just
integration into "the window of presence and global model of the
world", we'll see...

Does "Phenomenal self-presentation is anchored in mental
self-presentation" mean that what is experienced is always part of
what could be experienced? I.e. that there's always more to direct
attention to? It probably speaks about the world-model and self-model
structure.

"It is, of course, an interesting question, whether the abstract,
normally unconscious processing stages preceding volitional and
phenomenally self-modeled movement selection can already count as
egocentric representations, or whether this is precisely the step at
which those computations are integrated into a self-representation,
which also makes them conscious. In any case it now seems plausible to
assume that what gets integrated into the PSM of the organism as a now
deliberating subject is a determinate, single, and concrete
representation of a specific behavior."

The PMIR might be the key to intersubjectivity: its representation
can be its object.

Goal representations (via goal-encoded objects) are in nonegocentric
frames of reference, they can be integrated into the PSM for actual
behavior and for self-simulation, they can be integrated into
"allocentric frames", simulations of others integrated into PSM as
simulations, or just unconsciously activated by mirror neurons. And
mirror neurons first emulate low-level, non-goal movements. Low-level
and high-level resonance mechanisms do not coincide. Goal-encoded
objects are object representations with selection mechanisms for a
repertoire of actions like various grasping behaviors.

Linguistic concepts are much more than simple concepts (i.e. processes
that represent other processes non-homomorphically but by activation
links, and so gain recombinability), they are goal-encoded objects,
having qualia as all objects, here the words or other symbols.

piątek, 3 sierpnia 2012

Foundations of my systemic metaphysics

My inner anti-anti-realist: "Let's do some metaphysics!"
My inner positivist: "You must be kidding! Such frivolous activities will waste your mind."
My inner neopragmatist: "Come on, inner positivist. Some frivolity is not necessarily a bad thing. Would you rather have us watch 'Dark Knight Rises'?"
My inner positivist: "Well... I'm not taking part in this, anyway."
My inner neopragmatist: "I'd rather have us write some poetry, but I'm not gonna spend this evening alone. Bring out your toys, inner anti-anti-realist."

Reality is indexical

My inner anti-anti-realist claims that the following is roughly right, although it might require corrections.
  1. Objects of mathematical nontrivial theories are some of the processes: their processuality is the way in which they are constructed from the corresponding theories. Constructivists are invited to consider only constructive mathematics. Other processes are similar in that they are not "standalone", but are parts of systems.
    1. Finitists are encouraged to share their mathematics with constructivists and read below about what mathematical objects are real.
  2. We divide processes into two disjoint groups: platonic processes and physical processes. Platonic processes are mostly mathematical objects, but also qualia are platonic processes too; if the reader does not wish to think that qualia are real, then mathematical objects might suffice; if the reader has a bent towards humanities, then even literary characters can qualify as platonic processes. Platonic processes are "not fully individuated". Physical processes are "fully individuated". They are mostly the processes which could in principle have the reader as a necessary cause, and other processes like that. Even if the reader participates in a creative process, the platonic deed is "discovered", is a cause of the physical creative process.
    1. It is allowed for physical processes to in some sense be mathematical objects.
  3. A metaphysical system consists of a triple: a separation of processes into platonic and physical ones, a distinguished process (the index), and a ternary relation among processes, called causation.
    1. A metaphysical system serves as a formalization of the philosophical concept of reality.
  4. We call a direct cause the antecedent, first term of a causation relation, we call direct effect the consequent, second term of a causation relation, and we call the third term of a causation relation the individuation.
    1. A single process cannot have two direct effects with the same individuation.
    2. You can think of individuation as fixing of parameters or fixing of the reference frame.
    3. We are speaking in terms of necessary causes, not of sufficient causes.
    4. A process can have many direct causes and effects.
  5. We call a cause and an effect the antecedent and consequent of the transitive closure of the causality relation, ignoring the individuation term.
    1. When we talk about cause and effect but require the existence of individuation, it is a shorthand for talking about direct cause and direct effect.
  6. We define past reality as all causes of the distinguished process:
    1. The distinguished process is past-real.
    2. A cause of a past-real process is past-real.
    3. No other processes are past-real.
  7. If a process correlates with an object of mathematical theory, that object is one its causes.
    1. I have not yet established, how causation among mathematical objects, i.e. platonic processes, looks like.
  8. A platonic process cannot have a physical cause. But see the following point.
  9. If a physical process A has platonic cause B with individuation X, process B has platonic effect C, and physical process D has platonic cause C with the same individuation X, then A has effect D.
  10. Arbitrary one of the physical processes that correspond to the unity of the consciousness of the reader is the distinguished process (the index).
    1. Here the use of physicality forbids the "leakage" of reality, where only an abstraction of the reader is taken as an index.
  11. Every information flow is a causal flow.
  12. New information cannot appear from nothing: a  non-decomposable process adds information by combining causes. There has to be sufficient number of processes to account for all information.
  13. We define necessity inductively:
    1. The distinguished process is necessary.
    2. A cause of a necessary process is necessary.
    3. For a process P that has at least one direct cause that is necessary and is not an effect of P, take S to be the set of all processes that are both a cause and an effect of P (possibly indirect). If all causes of all processes in S are either necessary or in S, then P is necessary (and therefore S are necessary).
      1. No other processes are necessary.
    4. We call real all necessary physical processes and their causes.
    5. A possible universe is a set of processes U that is closed under necessity, i.e.:
      1. A cause of a process in U is in U.
      2. For a process P that has at least one direct cause that is in U and is not an effect of P, take S to be the set of all processes that are both a cause and an effect of P (possibly indirect). If all causes of all processes in S are either in U or in S, then P is in U (and therefore S is contained in U).
    6. A set S is a sufficient set of causes for a process C, if all causes of C can be traced back to S, i.e. if every possible universe that contains S also contains C.
    7. A possible universe whose all processes are platonic is called a platonic universe.
    8. simulation of a platonic universe U is a set of physical processes S, such that there is an individuation X, that for every platonic process B in U there is a process A in S such that B is a cause of A with individuation X.
      1. A synonym for "simulation" is "a physical correlate".
    9. When two simulations simulate the same platonic universe, we call one an emulation of the other.
    10. All necessary physical processes and their causes are real.
    11. An embedded metaphysics differs from a metaphysical system only in that point 11 does not hold for it.
    12. An embedded metaphysics A is valid with respect to a metaphysical system B via a homomorphism of causality h, if for every process P that is real in A, process h(P) is real in B.
    13. An embedded metaphysics A is possible with respect to a metaphysical system B via a homomorphism of causality h, if for every process P that is real in A, process h(P) is necessary in B.
    14. Materialism: were a real platonic process to differ, the physical processes over which it supervenes would differ as well.
      1. More generally: if only a subprocess of a process is essential for its effect, only the subprocess is a necessary cause of the effect.

    Exercise.

    Would the platonic "Permutation City" be real according to the physical characters of the book, if they used the presented metaphysics?