Archive: December, 2008

Philip Agre: Computation and Human Experience

[Readings] (12.15.08, 11:27 pm)

Philip Agre is a rare breed. He is a strong advocate of embodiment and situated action, and is also an AI practitioner. Agre was enormously influential on Michael Mateas, among others. Agre is interested in developing an approach to AI that is critically aware, as well as reformulating conventional AI applications in a manner that might be described as situated or embodied. His view on diectic entities has an enormous potential in application to an activity-centric approach to simulated characters.

Agre gives a good overview of the task at hand. His interest is in a change of approach around AI, based in both philosophical to technical arguments. He advocates the central idea of a critical technical practice. This idea has been very influential on Mateas and Sengers, in particular.


Agre’s goal is to shift from cognition to activity. AI has long been mentalistic, proclaiming to be a model of cognition and pure abstract thought. However, the practice of AI has tended to best work in application to specific practices and activities (note Hutchins). Theorem proving is not cognition in general, but it is a specific activity which can be computationally formulated and modeled. Focus on activity tends to model “situated, embodied agents.” (p. 4) The term “agent” applies to robots, insects, cats, and people, and has strength in ambiguity. Situation implies that an agent’s actions make sense within context of its particular situation. For an agent to be embodied, it must simply have a body: “It is in the world, among the world’s materials, and with other agents.”

There is a review and comparison of the planning based approach common to AI. Ordinary life has a sort of routine, and planning tends to make use of that routine. Agre poses that routines come from emergence. I would argue, though, that in social situations, aspects of routine may have been emergent, but they often become institutionalized. There are three questions that operate around activity, and each of these is important to consider for activity centered applications: (p. 7)

  1. Why does activity appear to be organized?
    Planning view: activity is organized because of plans.
    Alternative: orderliness is emergent. Form of activity is influenced by representations, but not determined by them.
  2. How do people engage in activity?
    Planning view: Activity is planned and contingency is marginal.
    Alternative: Activity is improvised, contingency is central. People continually redecide what to do.
  3. How does the world influence activity?
    Planning view: The world is fundamentally hostile. Rational action requires attempts to anticipate difficulties and everyday life requires constant problem solving.
    Alternative: The world is fundamentally benign. Environment and culture provides support for cognition. Life is a fabric of familiar activity.

AI work is itself a practice, which has its own values: Getting computational systems to work. Systems that do not work lack value. This lends to the idea that only things that can be built can be believed. On the other hand, building is also a way of knowing. Agre argues that emphasis should be shifted from technical product, to practice. The use of models and ideas is torn between science and engineering, both practices which led to the development of AI. Science aims to explain with models, whereas engineering seeks to build. AI straddles these two drives. Similarly, works of art exist on that same border, they are both explanations (expressions), and constructed products.

Metaphor in practice

There are two parts to Agre’s thesis: (1) AI is organized around the metaphor of internality and externality. The mind has an inside and an outside and discrete inputs and outputs at the interface. (2) A better starting point uses a metaphor of interaction, and focus on activity. The important part behind this is to be critically aware of the metaphors used in discourse about the practice.

There is an important nugget here: The technical criticism is with the lack of self reflection. A model is an interpretation and a language, it is a way of seeing things that is inherently metaphorical in nature. There is a double vocabulary, one at the level of discussing the subject of the model, as well as discussing the model itself and its interaction with the software. This duality of discourse is reminiscent of Mateas. “Scientific inquiries based on technical modeling should be guided by a proper understanding of the nature of models. A model is, before anything else, and interpretation of the phenomena it represents. Between the model and the putative reality is a research community engaged in a discursive operation, namely, glossing some concrete circumstances in a vocabulary that can be assimilated to certain bits of mathematics. The discourse within which this process takes place are not transparent pictures of reality; nor are they simply approximations of reality. On the contrary, such discourses have elaborate structures and are thoroughly metaphorical in nature. These discourses are not simply ways of speaking; they also help organize mediated ways of seeing.” (p. 46)

Machinery and dynamics

The two metaphorical theories operational in AI research are mentalism and interactionism. Newell and Simon’s GPS seems interaction oriented at onset, but shifts focus onto abstract representation very quickly. GPS aims for disembodied conceptions of the world. It equates objects with their representations, which is a dangerous phenomenological pitfall.

Agre proposes discussing computation in terms of machinery and dynamics. His goal is to discourage focus on machinery and instead emphasize dynamics. However, as presented, machinery is a useful metaphor. It implies situation within an environment: a machine has a physical presence, which steps away from raw functionalism. The term machinery is also very Deleuzian, which may be either positive or negative. Dynamics instead centers on interaction. Agre encourages us to get rid of computational machinery, and instead invent new dynamic effects, rather than devices.

Abstraction and implementation

Abstraction and implementation is an interesting dyad: Functional definition versus physical construction. What is the relationship between these to representation? It is important to discern the levels of abstraction at work when constructing artifacts. Newell, Simon, and Shaw constructed GPS which is not a model of cognition, but a model of problem solving in a particular domain. There is a generative element in theory, common to the work of both Newell and Simon as well as Chomsky. Soar bills itself as a “unified theory of cognition,” but is still primarily concerned with abstraction. Minsky is a researcher who seems to take this into account: he claimed to represent a constellation of mini-theories which are in turn heavily situated. This approach emphasizes implementation rather than abstraction.

Dependency maintenance

There is a review of critical technical practice. Agre’s aim is not to break the traditions of AI or start over, but to become critically aware of representation and computational machinery. The subject matter in Agre’s examples fold back to everyday activities. The goal is to see if the planning paradigm tells “reasonable stories” about everyday activity. Effectively, this means to see if the planning view of cognition reasonably accounts for everyday interactions. This involves a comparison of activity and routine.

  1. Different individuals use different routines for the same task.
  2. Routines emerge from repetition and practice.
  3. Novel activity is not routine.
  4. Routines are upset by circumstances.

Planning and improvisation

Hayes-Roth and Hayes-Roth: real planning is opportunistic and incremental. A relevant question about real world planning is this: “How is it that human activity can take account of the boundless variety of large and small contingencies that affect our everyday undertakings while still exhibiting an overall orderliness and coherence and remaining generally routine? In other words, how can flexible adaptation to specific situations be reconciled with the routine organization of activity?”

Running arguments

The situation and activity oriented framework described that is an alternative to planning is “Running Arguments” or RA. The cycle of RA is determined by the following steps: (p. 175)

  1. The world simulation updates itself.
  2. The periphery computes new values for the central system’s inputs, which represent perceptual information.
  3. The periphery compares the new input values with the old ones. Any newly IN input is declared a premise, any newly OUT input is declared no longer a premise.
  4. The central system propagates dependency values and runs rules. Both the dependency system and rule system continue to run until they have both settled.
  5. The periphery inspects the values of the central system’s outputs, which represent motor commands.
  6. The periphery and world simulation together arrive at as set of proprioceptive propositions (judgments about the success or failure of the agent’s primitive actions) and a set of motor effects (the immediate physical consequences of the agent’s actions).
  7. The world simulation updates itself again, and so on ad infinitum.

There is a set of principles regarding RA, discussed on (p. 179):

  1. It is best to know what you’re doing. Executing plans is derived from the symbols of the plan, not from an understanding of the situation. I think this suggests a model which makes use of the situation within the symbols themselves. Flexibility relies on understanding of situation and its consequences.
  2. You’re continually redeciding what to do. Much of the relevant input from the world is constantly changing, rather than remaining fixed.
  3. All activity is mostly routine. Activity is most frequently something which has been done before.

Representation and indexicality

Actions are about the world. This is an important idea! Aboutness ties into intentionality. At the same time, Agre argues that “world models are the epitome of mentalism. On its face, the idea seems implausible: a model of the whole world inside your head. The technical difficulties that arise are obvious enough at an intuitive level.” (p. 225) Knowledge representation is either mentalistic or platonic, and both reinforce the planning approach. Mentalistic models imply the existence of a world model in some memory state somewhere. Platonic models refer to nouns and items in purest abstract, appealing to universal qualities. Models of this nature resemble things like the Cyc project.

For a variety of reasons, I am very defensive of models, but world models have to do with an area that is extremely problematic, specifically knowledge representation. Understanding of an agent or character’s general knowledge, as well as knowledge about other characters and the world, are extremely difficult to model coherently. Existing theories, especially those around mental models, tend to use propositional models, which do not seem appropriate. Agre’s alternative is indexical representation, which is relevant for situational circumstances, but does not seem appropriate for larger scale activity. Agre’s alternative approach derives from phenomenology, especially Heidegger and Merleau-Ponty.

Diectic representation

Diectic representation is a different kind of world representation. The examples given are immediately situational: “the-door-I-am-opening, the-stop-light-I-am-approaching, the-envelope-I-am-opening, and the-page-I-am-turning.” (p. 243) These are diectic entities, which are indexical and functional. They are also immediately given a perspective, as all of these contain some reference to the entity itself. I would argue that these diectic entities absolutely tie into models, but those models are situational ones. There is considerable challenge to the idea of diectic entities as symbols (note Vera & Simon), but their use is very different from that of conventional symbols.

The most important distinction between diectic entities and symbols is their situated and non-objective nature. “A diectic ontology, then, is not objective, because entities are constituted in relation to an agent’s habitual forms of activity. But neither is it subjective. These forms of activity are not arbitrary; they are organized by a culture and fit together with a cultural way of organizing the material world.” (p. 244) This phrasing anchors diectic representations in a cultural basis. Given this context, it would make sense for diectic representations to be variables in defining activities in a simulated world populated by agents.


Mentalism ties into psychic unity (note Shore). “Perhaps the most unfortunate cultural consequence of mentalism is its tendency to collapse all of the disparate inside phenomena into a single ‘mind’ vaguely coextensive with the brain.” (p. 307) Mentalism too yields to objective accounts of reasoning, which proclaims a kind of universality of goal oriented thought characterized by Western philosophy.

Reading Info:
Author/EditorAgre, Philip
TitleComputation and Human Experience
Tagsai, specials, embodiment
LookupGoogle Scholar, Google Books, Amazon

Ortony, Clore, and Collins: The Cognitive Structure of Emotions

[Readings] (12.13.08, 7:20 pm)

Ortony, Clore, and Collins define a cognitive approach for looking at emotions. This theory is extremely useful for the project of modeling agents which can experience emotions. The cornerstone of their analysis is that emotions are “valenced reactions.” The authors do not describe events in a way that will cause emotions, but rather, emotions can occur as a result of how people understand events. This approach is surprisingly subtle and nuanced. There are many constraints and caveats, but these are all logical considering the perspective of the model.

The goal of this book is not to claim that the representation of emotion is exactly correct, but that the approach for thinking of them is cognitively viable. Emotional systems, as analyzed, are culturally dependent. There is a claim for generality, but not for universality. Emotions may be understood in terms of eliciting factors and valenced reactions. This is the heart of the theory. The actual emotions that result, including how they are thought of, and the words we use to describe them, are not claimed to be absolute, necessarily dependent on the factors, or universal.

It is extremely challenging to study emotions. There is a conflation between emotions and emotion words. The study here looks at enabling factors and conceptualized situations. The challenge in this study is to determine the relationship between cognition and emotion. Existing theories are: arousal/appraisal, and activation/valence. These theories account for (1) what the emotions are, and (2) their relative intensity. The existing approaches of study tend to return to emotion words, which are problematic because of their cultural and linguistic dependence. Words tie into systems of meaning that surround language, and rarely map one-to-one with emotions themselves. Instead, the authors arrive at the following definition: “Our working characterization views emotions as valenced reactions to events, agents, or objects, with their particular nature being determined by the way in which the eliciting situation is constructed.” (p. 13)

The Structure of the Theory

A question that is introduced are “what are basic types of emotions?” The authors view these as being grouped together by similar eliciting conditions. An idea that is introduced is to test the theory with computer models. In these cases, the computer programs are not thought to experience emotion, but rather, to be able to understand human emotions. There is a concern about the humanistic and phenomenological issue of understanding emotions. It may be argued that experience is necessary to understand emotion. This argument is reasonable, but does not seem to be a problem in the structure, given the way the reasoning works here. The theory describes reactions, valences, and eliciting conditions, but not the actual experiences themselves.

The formulation of the emotion types relate to how the world is understood in terms of agents, objects, and events. How emotions might emerge is very dependent on how individuals actually perceive and interpret the events. The authors give a somewhat distressing example of the emotions one might experience on learning that their neighbor is a child beater (p. 20). The person might think of the neighbor as an agent, which would give rise to reproach because of violation of standards. Thinking of the event might give rise to pity for the neighbor’s children. The person might consider the neighbor as an object, and experience hatred. This is a complex network of emotions that may arise from a relatively straightforward situation, but they may be accommodated within the model. Models in games and other forms of social simulations rarely address the multiple ways that the world may be perceived emotionally.

A predominant approach in cognitive science of emotions has been to view emotions as arising from a palette of “basic emotions.” This has been supported by many authors, including Oatley and Johnson-Laird. These perspectives tend to perceive basic emotions as “low level” feelings such as anger, disgust, fear, joy, sadness, surprise (from Ekman et al., 1982). These are chosen based on actions and behaviors, universal facial expressions, instincts, etc. The view of these emotions as basic is inconsistent with the theory of emotions as valenced reactions. They poke a few holes in the basic emotion theories, especially with the conflation of anger with distress, or anger with aggression, and avoidance with fear. Anger especially is given as a complex and joint reaction: as a combination of distress and reproach, unlike other theories which describe it as basic. There is also a conflation of emotions with mental states, although this is particular to the definition of emotion given by the authors.

Mental states can lead to valenced reactions, but are not necessarily reactions in of themselves. An example given by the authors is abandonment. The state of being abandoned is not an emotion until the individual reacts to that state. Furthermore, the type of reaction can be dramatically different depending on how the individual construes the reception of the state. Mental states can affect emotions, for instance, surprise is a state, but it tends to intensify emotions that react to the surprising event.

The Cognitive Psychology of Appraisal

This section is on appraisal, which is how situations are interpreted, so as to enable valenced reactions. Appraisal operates on three levels: motivation, standards, and attitudes. The authors first discuss appraisal as it connects to motivations (which are generally referred to as goals). Motivation is also a good term, due to the depth given by Maslow. Goals on their own tend to be rather shallow, dwelling on one thing at a time. The focus at the moment is on goals. The authors acknowledge that goals as employed in life tend to be spontaneous as opposed to planned. They suggest that a goal structure as might be imagined by a person is “virtual”, meaning that it is self-perceived. This gives some flexibility, enabling a certain freedom in how individuals might imagine their goals as to construe reactions to them. In this sense, goals are imagined. A virtual goal structure may be improperly formed (subgoals may not lead to completion of supergoals) and the goals themselves may not even be what the individual wants.

The virtual goal structure forms a graph reminiscent of planning systems. Nodes (goals) relate to each other by links that describe necessity, sufficiency, inhibition, and so on. The framework which is used later is borrowed from Schank and Abelson. The different goals are Active Pursuit, Interest, and Replenishment (A, I, R). An active pursuit goal would be something that the agent is engaged in pursuing, an interest goal is something that the agent wants to occur (it may even be something impossible for the agent to achieve on its own!), and a replenishment goal is something which must be renewed with some regularity (such as satisfying hunger). Appraisal of motivation and goals relates to the emotions which are reactions to events. These affect the emotional dimension of desirability. Goals are usually desired, but an agent may have goals for things not to occur, which would make those events be undesired.

Standards, like goals, may not necessarily be consistent. Standards tie into the appraisal of agents, who may conform with the observer’s standards or not. The degree of conforming ties into the emotional variable of approval. Standards are complicated to formulate, but very flexible, especially in consideration of cultural value systems. Some examples of standards given by the authors are “one ought to take care of other people’s things.” So, in fictional domains, for instance, characters might have wildly different morals and standards, and these would affect how characters approve or disapprove of each other. Events which bear on standards can also be reacted to on their own, and yield event related emotions. Attitudes are much more obtuse, and relate to personal tastes. Attitudes affect the attraction emotions, which focus on objects rather than events or agents.

Desirability ties into goals, but also expectations. Appraisal of an event is done in perspective. The authors give an event of an IRS refund of $100, which would be desirable if one expected nothing or to owe money, but would be less desirable than a refund of $300. Praiseworthiness relates to standards, and because it relates to agents, a central variable is responsibility. An agent must be considered responsible in order to be accountable to standards. This perception of responsibility may not be rational (such as blaming the dog that ate your birthday cake, or the computer program that erased your paper), but it is consistent with our understanding of intentionality.

Appealingness reflects attitudes. An attitude is a disposition, which is not an emotion alone. One may have a disposition to like ice cream, but this does not indicate an emotion on its own, rather it indicates the potential for an emotion when ice cream is present. The relation between attitudes, dispositions, and emotions are difficult to express in language because common usage tends to conflate the concepts.

The Intensity of Emotions

There are 4 global variables that affect the intensity of emotion. Local variables are ones that affect the individual emotions themselves. Global variables affect all emotions that one might experience at a given moment. The authors divide between local and global intensity variables based on isolatability. Variables ought to be independent and not modulate each other.

  1. Sense of reality. When the eliciting conditions are perceived as real, the emotions are more intense. This has to do with both literal understanding of reality (the plane might be delayed, versus the plane is delayed), as well as a sense of investment. If an aspiring author wants to write a best-selling novel, then the emotions associated with the success of the novel will be relative to whether the author considers the project to be realistic or a fantasy.
  2. Proximity (psychological). Proximity relates to factors such as time, so emotions pertaining to remembered events are less strong than the experience of the events. Proximity also relates to psychological nearness, as relates to reactions to consequences for others. Proximity can be an issue in the case of reactions to tragedies in far away places, or good or bad things happening to strangers.
  3. Unexpectedness. Unexpectedness relates to both likelihood and suddenness. Reactions to a sudden catastrophe is more intense than one that was forseen (although the emotions themselves may change from shock to self-reproach). The disappointment at losing at the lottery is less severe because loss is expected, than say being rejected from a job application which seemed very likely. The intensity due to unexpectedness has much to do with the perceived normality of the situation.
  4. Arousal. Arousal is the total experience of emotions that one has experienced over a period of time. Gradually, arousal dissipates, but successive reactions can increase arousal, which will in turn, increase the intensity of subsequent reactions. The authors give an example of someone who is preparing breakfast for his family, but everything goes wrong. He burns the toast, forgets to start the coffee soon enough, overcooks the eggs, and so on. These events lead to reactions which ultimately cause a great deal of frustration. He might slam cabinet doors or speak rudely to people. The situation may be attributed as events, or as agents (either the cook or the appliances are agents which are behaving irresponsibly), and these reactions may activate scripts about personal failure and inadequacy.

The chief local variables are desirability, praiseworthiness, and appealingness, as discussed in the previous section. The authors introduce further local variables that pertain to other emotion types:

  1. Desirability: event based emotions. (pleased/displeased)
  2. Praiseworthiness: attribution emotions. (approving/disapproving)
  3. Appealingness: attraction emotions. (liking/disliking)
  4. Desirability for other: fortunes of others. Whether the event is desirable for the other.
  5. Deservingness: fortunes of others. Whether the other “deserves” the event.
  6. Liking: fortunes of others. Whether the other is liked or not. These distinguish between: happy-for, pity, gloating (schadenfreude), and resentment.
  7. Likelihood: prospect emotions. (hope/fear)
  8. Effort: prospect emotions. How much effort the individual invested in the outcome.
  9. Realization: prospect emotions. The actual resulting outcome. These distinguish between: relief, disappointment, satisfaction, and fears-confirmed.
  10. Strength of identification: attribution emotions. The stronger one identifies with the other, that distinguishes between whether pride or admiration is felt.
  11. Expectation of deviation: attribution emotions. Distinguishes whether the other is expected to act in the manner deserving of admiration or reproach. These distinguish between: pride, shame, admiration, reproach.
  12. Familiarity: attraction emotions. (love/hate)

Reactions to Events: I

The authors give a fine grain analysis that explores specific context of emotion instances. Loss is a specific instance of distress in this typology, and other emotions, that are kinds of losses derive accordingly. Grief is loss of a loved one, homesick is loss of the comforts of home, loneliness is loss of social contact, lovesick is loss of the object of romantic love, regret is loss of opportunity (p. 91). These may be considered different emotions in the sense of emotion words, but derive from the same kinds of experiences, the same kinds of reactions.

Reactions to Events: II

The authors give a table of relative outcomes matched with prospects and expectations. This lends to a mix of prospect and well being emotions. (p. 129)

Prospect Outcome Emotions
-$1000 -$1400 fears confirmed ($1000)
unexpected distress ($400)
-$1000 fears confirmed ($1000)
-$400 relief ($600)
fears confirmed ($400)
$0 relief($1000)
+$400 relief($1000)
unexpected joy ($400)
+$1000 +$1400 satisfaction ($1000)
unexpected joy ($100)
+$1000 satisfaction ($1000)
+$400 disappointment ($600)
satisfaction ($400)
$0 disappointment ($1000)
-$400 disappointment ($1000)
unexpected distress ($400)
Reading Info:
Author/EditorOrtony, A.; Clore, Gerald; Collins, Allan
TitleThe Cognitive Structure of Emotion
Tagssocial simulation, ai, specials
LookupGoogle Scholar, Google Books, Amazon

Newell and Simon: Human Problem Solving

[Readings] (12.12.08, 5:09 pm)

Newell and Simon are enormously influential in AI and cognitive science. Their approach to human problem solving has however been a favorite target for criticism among advocates for embodiment, situated action, and activity based approaches to cognition. I have been finding myself taking up the mantle of this group, so I am approaching Newell and Simon from a critical perspective. It is not my aim to reject symbolic approaches to problems and problem solving, but rather to change how symbols are conceived of and used.

This is the text that introduces the General Problem Solver, or GPS. GPS eventually was transformed into Soar, which remains a usable problem solver.


Early on, the authors introduce the aim of the book as to develop a theory of cognition. Cognition is framed as a combination of three elements, which are performance, learning, and development. The emphasis in this book is performance, which is how cognition is used. The study of the book is introduced very early: “The present study is concerned with the performance of intelligent adults in our own culture. The tasks discussed are short (half-hour), moderately difficult problems of a symbolic nature. The three main tasks we use–chess, symbolic logic, and algebra-like puzzles (called cryptarithmetic puzzles)–typify this class of problems. The study is concerned with the integrated activities that constitute problem solving. It is not centrally concerned with perception, motor skill, or what are called personality variables.” (p. 3-4) This isolation of the problem is very specific and cuts out a huge class of cognition and activity. We should take Newell and Simon’s conclusions from this study as being applied specifically to this domain of problems, but it often gets conflated with something much greater. Performance and problem solving of chess, logic, and cryptarithmetic puzzles are not at all representative of cognition on the whole.

Performance on problems is compared to learning. Learning is described as a secondary activity, which is primarily dependent on performance. This type of reasoning is directly at odds with work by Vygotsky and Tomassello and other developmental psychologists. The study also omits “hot cognition,” which is cognition that is emotional, reactive, or dependent on personality. These forms of cognition are arguably much more embodied than the “cold” problem solving described by the problem domain. The authors argue that hot cognition should follow and derive from problem solving. This is emblematic of a trend in this form of thought, that other more tricky areas of cognition can just be added on to the architecture.

The study can also be viewed as a theory of psychology. Unlike other forms of theories, this one is content oriented (as opposed to process-oriented). Here the authors explain that a content oriented theory can serve as a model for the content: “The present theory is oriented strongly to content. This is dramatized in the peculiarity that the theory performs the task that it explains. That is, a good information processing theory of a good human chess player can play good chess; a good theory of how humans create novels will create novels; a good theory of how children read will likewise read and understand.” (p. 10-11) What is described here seems to be a generative model, as opposed to an analytic model. However, this authors’ claim is fraught with three counterexamples. Chess programs approach a game of chess very differently than chess players do. Algorithms for story generation (to say nothing of whole novel generation) are leagues behind actual human writers. And text understanding systems are also fraught with difficulty (notably the Bar-Hillel problem for word contextualization).

There is another more fundamental critique with the claim that a content oriented theory will perform the task in question. The issue is that a theory does not actually practice or perform the task that humans do, it merely models it. AI does not think, but it models thinking. It does not solve problems, but solves models of those problems. This distinction is subtle but important. If a problem is easily and non-problematically transformable into a symbolic model, then the solved model should solve the actual problem, but this is only as good as the transformation of reality into the model. Someone may program a computer to solve a model of a problem, such as a disease outbreak, but if the model in the program omits crucial details that arise in the complexity of real life, then the solution is intrinsically flawed.

It is relevant to note the authors’ use of the word “dramatized” in the citation. Dramatization can take on an interesting meaning here when compared to the elements of performance, resemblance, and enactment. A wholly legitimate way to examine AI and problem solving is that the computer dramatically enacts the problem. That is, it represents the problem and the solution dramatically, in the sense of performing to the user meaningfully.

Task Environments

Problem solving takes place in a task environment. The task environment contains the symbolic content necessary to solve the problem. The authors propose using the model of the rational economic man (who maximizes utility). The task (and hence utility) are defined by the task environment: “The term task environment, as we shall use it, refers to an environment coupled with a goal, problem, or task–the one for which the motivation of the subject is presumed.” (p. 55) Performance is inextricably tied to rationality, and the situation is transformed to the task environment. Task environments are internally represented. The internal representation is a symbolic construction which illustrates the problem space: all moves, states, and outcomes.

Internal representations are tied to linguistic structure (which is inherently semiotic). For transcribing problem solving into an information processing system, objects are mapped to symbols. This is unproblematic in the examples given (chess moves, symbolic logic problems), but is much more troublesome in other situations. Even in the example of a chessboard, there are many ways to look at the positions of pieces and encode them symbolically. For example, a pattern would probably be treated as a symbol by a grandmaster. It is obvious to represent piece coordinates as symbols, but it is not the only approach. In domains where the units are less obvious, the symbolic transformation is even more fraught with complexity.

Newell and Simon give their definition for a problem: “A person is confronted with a problem when he wants something and does not know immediately what series of actions he can perform to get it.” (p. 72) This definition alone is rather interesting. Here, a problem is tied to desire as well as possession and attainment. These are very embodied qualities. Given this definition, it is far from obvious what it would mean for a computer to be confronted with a problem. The answer the authors give is that problems can be represented using symbols, as constructed from set or 1st order logic.

Problem Solving

The problem solving process given resembles the architecture of Soar. (p. 88)

  1. An initial process, here called the input translation, produces inside the problem solver an internal representation of the external environment, at the same time selecting a problem space. The problem solving then proceeds in the framework of the internal representation thus produced–a representation that may render solutions obvious, obscure, or perhaps unattainable.
  2. Once a problem is represented internally, the system responds by selecting a particular problem solving method. A method is a process that bears some rational relation to attaining a problem solution, as formulated and seen in terms of the internal representation.
  3. The selected method is applied: which is to say, it comes to control the behavior, both internal and external, of the problem solver. At any moment, as the outcome either of processes incorporated in the method itself or of more general processes that monitor its application, the execution of the method may be halted.
  4. When a method is terminated, three options are open to the problem solver: (a) another method may be attempted, (b) a different internal representation may be selected and the problem reformulated, or (c) the attempt to solve the problem may be abandoned.
  5. During its operation, a method may produce new problems–i.e., subgoals–and the problem solver may elect to attempt one of these. The problem solver may also have the option of setting aside a new subgoal, continuing instead with another branch of the original method.

Logic: GPS and Human Behavior

Work here shows students solving problems and their actions written in the formal logic of GPS. Experimental subjects are either Yale or CMU students. The question of what the students are doing seems to trace back to their formal logic training. The problem domain is still a highly symbolic one to begin with. The problem solving approach is thus not really a representation of how humans solve problems, but how problems tend to be solved within the formal structure of this problem domain.

The Theory of Human Problem Solving

This section is where GPS (and Information Processing Systems in general) are translated into human problem solving. Obviously, my scruples are with the assumptions. The propositions which pose the questions for human problem solving (the assumptions) are these: (p. 788)

  1. Humans, when engaged in problem solving in the kinds of tasks we have considered, are representable as information processing systems.
  2. This representation can be carried to great detail with fidelity in any specific instance of person or task.
  3. Substantial subject differences exist among programs, which are not simply parametric variations but involve differences of structure and content.
  4. The task environment (plus the intelligence of the problem solver) determines to a large extent the behavior of the problem solver, independently of the detailed internal structure of his information processing system.

The propositions to be answered by the chapter are as follows: (p. 788-789)

  1. A few, and only a few, gross characteristics of the human IPS are invariant over task and problem solver.
  2. These characteristics are sufficient to determine that a task environment is represented (in the IPS) as a problem space, and that problem solving takes place in a problem space.
  3. The structure of the task environment determines the possible structures of the problem space.
  4. The structure of the problem space determines the possible programs that can be used for problem solving.

External memory is discussed, but affordances and cognitive extension into situation or environment are not discussed. This has a very mentalistic approach, where symbolic structure of environment is not addressed. The only symbols are in the problem space itself.

Reading Info:
Author/EditorNewell, Allen and Simon, Herbert
TitleHuman Problem Solving
ContextThis is the canonical work that introduces GPS and they symbolic approach to problem solving
Tagsspecials, ai
LookupGoogle Scholar, Google Books, Amazon


[General] (12.12.08, 12:36 pm)

I’m updating to WordPress 2.7. Hopefully this will not break anything. In case it does, it will likely take a little while for it to get fixed. Just a heads-up.

Mathematical observations

[Experiments,General] (12.11.08, 5:16 pm)

So, I spent some time fiddling around with the Henon applet. It’s possible to discover and get a feel for several properties of the map just through experimenting, and that’s a really positive sign. You can easily identify the two fixed points, and it’s possible to tell what their relative stabilities are. It’s also possible to make some other interesting visual observations.

One thing that stands out in mind is a way in which the Henon map is different from, let’s say, a Julia set. The filled in gray region represents the space where the orbit of every point in the set does not diverge. Generally, points in this region will fall into the characteristic parabola saddle shape shown by the white points. This means that points in this region have a chaotic orbit. Points in this region that are somewhat apart will eventually be separated no matter how close they start together. However, if you choose any point on the border of the gray region, it will converge to the unstable fixed point on the left of the map. Even though this point is unstable along one axis, it is very compressed along the other. I don’t have a proof for this, but it seems visually evident.

What is interesting about this, is that the situation here is exactly the opposite of Julia sets in the canonical z->z^2+c map. In these cases, the border of the filled-in Julia set is chaotic, any two points on this border will eventually become separated. Whereas points within the filled-in set will always converge to some cycle.

It seems, given this, that the filled-in region is actually the Henon map’s Julia set, whereas the border is the “filled-in” Julia set. Or maybe there’s a different term for it. It’s been a while since I’ve done math, so it’s hard to know.

Fun with the Henon map

[General,Toys] (12.10.08, 11:08 pm)

I just built something that I’ve been meaning to make for years. I tried building something like it before, about 5 years ago, right before I graduated from CMU. At the time, I had strong programming ability for small projects, but didn’t really know how to write programs modularly with any effectiveness. I’ve learned so much since then, so now something that took a really inordinate amount of time and never got off the ground before took two days to write. I like to program recreationally. It’s a very bizarre habit. It’s not a compulsion, but really just a passtime.

The project in question is a visualizer for strange attractors. I did research on them as an undergraduate at CMU, and I’ve been wanting to make this project as a general tool ever since. I wanted something that could display stuff in both parameter space and phase space, and effectively get at all of the peculiar things that can happen with strange attractors.  The example below only does anything interesting in phase space, but it’s rather flexible, and has a nice modular architecture, which means that it will be easy to adjust properties, add or remove visual elements, and generally do interesting things.

It does not seem to behave quite properly with the mouse wheel at the moment, but it should have zoom functionality…

Java 1.5 or higher is required to run this applet. Please download a JRE from java.sun.com.

Further adventures in paper writing

[General] (12.08.08, 11:58 pm)

I have been working on a paper for my cognitive science course, and that has been going well, but not quickly. The paper was turned in while somewhat incomplete. I will post it when it is done.

While I have not yet found a stable paper writing system yet, I have made progress. I am leaning toward Docbook for final products. It is intended for books rather than papers, which is an irritating flaw, but it does come with a large set of XSL transformations to transform a docbook document into a variety of formats. Its export to HTML is promising: sections are all classified with “class” attributes, so transformed documents can be very easily styled. For the daily practice of writing, I’m happily chugging along with Google Docs.

When I say writing system, I mean something that handles writing, document formatting/typesetting, and dealing effectively with citations. While I have previously adored LaTex and BibTeX, the formats are very ill supported. If anyone ever finds a LaTeX editor that isn’t terrible, please let me know. While I have previously been unimpressed with Zotero, it is pretty effective as a reference manager. With references, I want something that I can have work with my bibliography, but that’s really dependent on the openness of the reference format and my ability to get my homebrew readings plugin to behave nicely.

Eric Mueller: Commonsense Reasoning

[Readings] (12.03.08, 3:44 pm)

This book is about an approach to commonsense reasoning that is implemented using formal logic. Mueller does not make the claim that humans make use of formal logic structures in understanding the physical world, but rather that this is a way for commonsense physical phenomena to be understood computationally. The approach does not replicate or even simulate the human method of cognition or understanding, but rather gives a formal account of relationships. Commonsense systems could be applied towards making predictions, performing diagnostics, or analysis within a simulated or artificial environment. Mueller uses classical first-order logic, and every element of commonsense reasoning takes the form of axioms that work within that space.

Mueller lays out a set of four assumptions that guide the rest of the investigation. These assumptions are certainly contestible, but they can be seen as a basis on which the rest of the work can be derived. (p. xx)

  • I assume, along with most cognitive scientists, that commonsense reasoning involves the use of representations and computational processes that operate on those representations.
  • I assume along with researchers in symbolic artificial intelligence, that the representations are symbolic.
  • I assume, along with researchers in logic-based artificial intelligence, that commonsense knowledge is best represented declaratively rather than procedurally.
  • I use the declarative language of many-sorted first order logic.

The formulation of commonsense reasoning accounts for a number of specific properties of real world objects. This contains the vocabulary and the elements that will be formalized into axioms of the commonsense reasoning logic. One noteworthy flaw within this is the element of perspective. For someone to have common sense, that individual must have a perspective. Mathematical logic in abstract does not have a perspective that is readily apparent. Even thought this logic can make commonsense conclusions, it still has a view from nowhere. (p. 7-8)

  1. Representation. The method must represent scenarios in the world and must represent commonsense knowledge about the world.
  2. Commonsense entities. The method must represent objects, agents, time varying properties, events, and time.
  3. Commonsense domains. The method must represent and reason about time, space, and mental states. The method must deal with object identity.
  4. Commonsense phenomena. The method must address the commonsense law of inertia, release from the commonsense law of inertia, concurrent events with cumulative and canceling effects, context-sensitive effects, continuous change, delayed effects, indirect effects, nondetermnistic effects, preconditions and triggered events.
  5. Reasoning. The method must specify processes for reasoning using representations of scenarios and representations of commonsense knowledge. The method must support default reasoning, temporal projection, abduction, and postdiction.

Commonsense logic originated with situation calculus as created by John McCarthy and Patrick Hayes in the 1960s. This was the inspiration for Robert Kowalski and Marek Sergot to develop event calculus, which is the method of Mueller’s investigation. The foundation of event calculus relies on the understanding of events and properties that change over time. Time dependent properties are called fluents, which could be typed variables or true-false values. Events are any occurrence that can happen within the world. Time in the event calculus is linear, and can be either discrete or continuous.

There are four main predicates that work on the event calculus (p. 11). Note that these are not elements of a “new” logical structure, but are rather constructions within the framework of first order logic.

  1. HoldsAt(f, t) represents that fluent f is true at timepoint t.
  2. Happens(e, t) represents that event e occurs at timepoint t.
  3. Initiates(e, f, t) represents that, if event e occurs at timepoint t, then fluent f will be true after t.
  4. Terminates(e, f, t) represents that, if event e occurs at timepoint t, then fluent f will be false after t.

Using these elements, many axioms can be declared. These define the ordering of time, how effects may be defined or triggered, how events may be preconditions, what cumulative effects are, what abnormal states are, and so on. These are elaborated in detail throughout the book. They work together to form a representation of commonsense reasoning within a domain. In addition to these axioms, a domain must include observations of the world’s properties at various times, and a narrative of the known events in the world (p. 35). Mueller gives a formal definition for a domain description (p. 37):

  • Positive effect axioms, negative effect axioms, release axioms, effect constraints, positive cumulative effect axioms, and negative cumulative effect axioms.
  • Event occurrence formulas, temporal ordering formulas (the narrative).
  • Trigger axioms, causal constraints, and disjunctive event axioms.
  • Cancellation axioms.
  • Unique names axioms.
  • State constraints, action precondition axioms, and event occurrence constraints.
  • Trajectory and antitrajectory axioms.
  • Observations.
  • Event calculus axioms.

This is an extensive list, but composed together it represents what would completely define a domain within the calculus. This is actually quite contestible when compared with human reasoning. Human reasoning is necessarily incomplete, and much of the logical formulations are never explicit. For the purposes of logical reasoning, it still seems somewhat rigid and inflexible. It is severely dependent on total objective knowledge. With missing or incorrect knowledge, the logic might be crippled.

Later on, Mueller gives a chapter on the Mental States of Agents. This too gives an external and objective perspective on the modeled phenomena. It necessarily has an external omnitient view inside the minds of the emotional agents. The ostensible goal of this is to develop a system which can reason about emotions, but that too depends on issues of understanding and perception. Agents themselves, as modeled, have beliefs, and thus are subject to some perspective, but the logic will reason about those beliefs without perspective.

Mueller first gives a version of the Beliefs, Desires, and Intentions framework (listed as Beliefs, Goals, and Plans). This gives a clear logical account for conclusions that may be derived from BDI agents and environments. This reasoning is still very complex, but could be made more rapid though computational implementation.

Next, there is a logical formulation of the emotion theory developed by Ortony, Clore, and Collins. The formulation uses the system of eliciting conditions. The goal is to make logical conclusions about the emotions of agents when events occur. This work creates definitions for the predicates, Joy(a, e), meaning that agent a is joyful about event e, and goes to more complicated predicates such as Appreciation(a1, a2, e), which means that agent a1 is appreciative of agent a2 for performing action e. Following this is a large series of axioms which formally defines the relationships between the various predicates of deirability, belief, joy, distress, hope, resentment, and so on.

Default reasoning is constructed using vanilla 1st order logic, not with any messy nonmonotonic logic, or probabilistic reasoning. The perspective here relies on a total account of abnormal conditions, and a total knowledge of the states of objects. This formulation is especially problematic from a sociological perpective because of its emphasis on the cases of normality. An example given is that apples are red, unless some abnormal condition applies, such as that the apple is a Granny Smith, or is rotten. Of course, the claim of normal or default conditions is highly contestible, and an architecture that encourages defaults could lead to problematic assumptions.

Reading Info:
Author/EditorMueller, Eric
TitleCommonsense Reasoning
ContextMuller describes event calculus, which can be used for describing states and knowledge
Tagsai, specials
LookupGoogle Scholar, Google Books, Amazon
« Previous Page