Harman, Gilbert; 1986. Change in View. MIT Press, Cambridge, MA.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....understanding [Grosz Sidner, 1990] and multiagent systems [Singh, 1991c] Perhaps the salient property of future directed intentions is that they involve commitment on the part of agents. This view has been gaining ground in the philosophical and AI literatures recently [Bratman, 1987, ch. 2] [Harman, 1986, p. 94] Cohen Levesque, 1990, p. 217] The idea here is that an agent who has an intention is in some way committed to it not only does he intend to achieve the relevant condition right now, but would also intend to achieve it later, even as the circumstances changed, perhaps for the worse. ....

Harman, Gilbert; 1986. Change in View. MIT Press, Cambridge, MA.

....model of the situation. From both a formal perspective and a philosophical perspective, two general prescriptive principles for guiding belief revision have been offered: minimal change and epistemic entrenchment. Minimal change is essentially a conservation of information viewpoint (see Harman, 1986). When faced with inconsistency, this principle advocates that a reasoner should formulate a new belief state that retains as much as possible from the prior belief state, while still removing the inconsistency. The problem with minimal change is that it is hard to quantify, and even when it is ....

....some beliefs seem to be more deserving of retention than others in the face of contradiction (Grdenfors, 1988) Some of this deserving quality may be grounded in a sort of utility viewpoint, under which some beliefs have a greater explanatory power than others. Doyle 1991; Grdenfors, 1988; Harman, 1986). Elements of this perspective have been realized in proposals for computing coherence of explanations (e.g. Thagard, 1989; 1992; Ng Mooney, 1990) But as Grdenfors (1993) notes, with the exception of expected utility, there has been little cognitive research concerning how expectations about ....

[Article contains additional citation context not shown here]

Harman, G. (1986). Change in view. Cambridge, MA: MIT Press.

....while believing that ff is not possible. This requirement on the agent s rationality can be captured by the following. 14 For an overview of BDI theories see the volume edited by O Hare and Jennings [1996] 15 Many of the important questions in this area were originally raised by philosophers [Harman, 1986; Davidson, 1989c; Bratman, 1987; Anscombe, 1963] they have since become of interest to many in AI as well [Cohen et al. 1990] The suggestion that the relation between mental states and action is causal has, in fact, been disputed by some philosophers [Anscombe, 1963] 22 Rationality Postulate ....

Harman, Gilbert 1986. Change in View. MIT Press.

....the agent observes the actions of its opponents in negotiation. The revision is done in the context of the equilibrium strategies and is based on the actions the opponents are supposed to take according to these strategies. 9 Belief revision has more recently been treated by philosophers (e.g. [49, 42, 60, 24, 22]) theoretical computer scientists (e.g. 20] and artificial intelligence researchers (e.g. 14, 64, 50, 27, 15] These groups view an agent s beliefs as a set of assertions (without probabilities) and revising beliefs involves deciding how that set of assertions should change when new ....

G. Harman. Change in View. MIT Press, Cambridge, MA, 1986.

....so that an operational semantics can be designed. For more formal treatments of these issues the reader is referred to [4, 6, 7, 23] I will follow Bratman [4] in defining intentions as a separate and identifiable mental state. This can be contrasted with the belief desire model of intentions [8, 17], in which intentions are not atomic, but are composed out of beliefs and desires. Unlike Bratman, who uniformly uses the term intention to denote any commitment toward an action (positive or negative) I have found it helpful to define three possible intentional states: positive, negative, and ....

Gilbert Harman. Change in View. MIT Press, 1986.

....[7, p. 94] invoke when they say that an agent is committed to his intentions. This is also the notion used in [2, p. 217] and studied in [12] As I will explain shortly, this sense of commitment also arises under the guise of the epistemic entrenchment of an agent s beliefs as in [4] and [7]. I call the first kind of commitment social commitment or S commitment and the second kind psychological commitment or P commitment. These concepts are the subject of this paper. The question: can a group be the locus of believing and intending may be raised. The underlying assumption of this ....

Gilbert Harman. Change in View. MIT Press, Cambridge, MA, 1986.

....may or may not be said to have an intention [ Cohen and Levesque, 1990, McDermott, 1982, Singh, 1990, Singh and Asher, 1990 ] The modern philosophical view is that intentions cannot be reduced to desires and beliefs (e.g. see [ Brand, 1984, pp. 121 125 ] Bratman, 1987, pp. 18 23 ] and [ Harman, 1986, pp. 78 79 ] Intentions are most often seen as being mutually consistent, compatible with beliefs, and direct or immediate causes of action (e.g. Brand, 1984, p. 46 ] This is a useful property for the purposes of this paper, since it helps relate intentions to rationality via actions. ....

....ones. It is the latter that will interest us here. Perhaps the salient property of future directed intentions is that they involve commitment on the part of agents. This view has been gaining ground in the philosophical and AI literatures recently (e.g. see [ Bratman, 1987, ch. 2 ] Harman, 1986, p. 94 ] and [ Cohen and Levesque, 1990, p. 217 ] The idea here is that an agent who has an intention is in some way committed to it not only does he intend to achieve the relevant condition right now, but would also intend to achieve it later, even as the circumstances changed, perhaps for ....

Gilbert Harman. Change in View. MIT Press, Cambridge, MA, 1986.

....from desires (which may be mutually inconsistent or incompatible with beliefs, and may not lead to actions) and beliefs (which do not in themselves lead to action) This view is supported by a number of philosophers, e.g. Brand, 1984, pp. 121 125 ] Bratman, 1987, pp. 18 23 ] and [ Harman, 1986, pp. 78 79 ] We restrict ourselves to intentions that are future directed, i.e. geared toward future actions or conditions. The literature over the past decade or so agrees on the idea that intentions involve some commitment on part of the given agent. This commitment is psychological ....

Harman, Gilbert; 1986. Change in View. MIT Press, Cambridge, MA.

....the consequences obtained from each source of information as in (Dubois et al. 1992b) rather than combining the knowledge bases attached to each source before the inference process takes place. Our dichotomy coherence versus foundation is somewhat different from the one used in the literature (Harman, 1986), G rdenfors, 1990) Rao Foo, 1989) Doyle, 1992) DelVal, 1994) In this paper, we do not assume any particular structure on the beliefs in the knowledge base (contrary for example to RMS defined in (Doyle, 1992) Nor do we assume any (in)dependence relations between beliefs. Moreover, ....

G. Harman (1986) Change in View, Principles of Reasoning.

.... changes from attending one theory of the world (scenario) to another) That seems to be either the role of a psychological theory (e.g. How many scenarios do people consider at once How many scenarios do people consider at all How much evidence is required before someone changes their mind [Harman86] or an implementation decision (e.g. Should we build one theory at a time and undo relevant assumptions if we get into trouble [Doyle79] or should we try to build all explanations or extensions at once [de Kleer86] Both of these are very important issues but are not the subject of this ....

G. Harman, Change in View, MIT Press.

....where an agent makes a commitment to others to do or not do certain actions. The agent is liable for not acting up on them. The other sense of commitment involves an agent by himself. The agent is committed to his intentions or beliefs, but is not liable (to anyone else) Bratman, 1987, ch. 2] [Harman, 1986, p. 94] Cohen Levesque, 1990, p. 217] Singh, 1991b] This is related to epistemic entrenchment [G ardenfors, 1988] We call the first kind of commitment S commitment (for social) and the second kind P commitment (for psychological) Historically, commitments were of the psychological, ....

....concept. Perhaps their salient property is that they involve a commitment on the part of agents. Intentions are taken to be causes of actions, and can involve actions in the future. This requires some persistence or commitment. This has been supported by several researchers [Bratman, 1987, ch. 2] [Harman, 1986, p. 94] Pollack, 1992] and [Cohen Levesque, 1990, p. 217] An agent who has an intention is in some way committed to it not only does he intend to achieve the relevant condition right now, but would also intend to achieve it later, even as the circumstances changed, perhaps for the worse. ....

Harman, Gilbert; 1986. Change in View. MIT Press, Cambridge, MA.

....relations. The amount of derivational information contained in the coherence relations would affect the behaviour of the system and how closely the revisions would match those of a purely foundations system. Harman takes another view of the compatibility of foundations and coherence theories [Har86] He puts forth that our beliefs actually fall into two categories: working hypotheses and accepted beliefs. The Principle of Conservatism which guides coherence theory is applicable only to accepted beliefs, Harman argues, and is necessary for those beliefs because of the bookkeeping involved in ....

....in which this would be desirable include user modeling and multi agent problem solving (both adversarial and co operative) Tha92] Second, people are very good at overcoming the problems noted in Section 1.1. People revise vast stores of beliefs frequently, efficiently, and effectively [Har86, KST82] People s solutions may not be optimal people do make mistakes but they are near enough to optimal to be a survival trait. 2.1 Types of Beliefs In [QU78] Quine examines the nature of beliefs. He breaks beliefs down into six major classes, depending on their source: ....

[Article contains additional citation context not shown here]

Gilbert Harman. Change in View. MIT Press, Cambridge, Massachusetts, 1986.

....To take a slightly less extreme view, one might accord observational reports (say) the status of full belief, but still no conclusions drawn from these would be certain. Presumably, there are certain computational advantages to be gained by ruling out possibilities that are very unlikely [10, 17]. Chief among these is the ability to exploit logical rules of inference. Such rules allow conclusions to be reached in manner that is independent of context, in contrast to probabilistic inference. The locality of logical rules can be exploited if parts of the belief are (treated as if they are) ....

Gilbert Harman. Change in View. MIT Press, Cambridge, 1986.

....contains n atomic formulas, Omega will contain 2 n elements, and require that many judgments on the part of an agent who wants a point valued probability. Without simplifications (for example a principle of indifference) this is an intractable problem. This is also pointed out by Harman, in [10]. Thus there is indeed a virtue in the interval representation from a purely descriptive point of view. Given a finite, feasible, amount of inquiry, we can determine the beliefs of an ideally rational coherent agent only up to a lower probability, which yields an upper probability through the ....

G. Harman. Change in View. MIT Press, Cambridge, MA, 1986.

....akin to Snell s law for computational systems: they propose taking the path of least resistance in the space of representations and computations. The representational irrelevance principle is a variation of Quine s principle of indiscernibility of identicals [Qui63] It is re echoed in Harman s [Har86] principle of clutter avoidance and Lenat s notion of cognitive economy [LHRK79] Whereas the representational irrelevance principle outlaws making unnecessary distinctions, the computational irrelevance principle outlaws unneeded computation. To apply the computational irrelevance principle we ....

G. Harman. Change in View. MIT Press, Cambridge, MA, 1986.

....in a theory by making the objects assumed by it as few and as large as is consistent with the correctness constraints. This ontological economy in describing problems entails computational savings. We only make distinctions necessary for the purpose at hand. This informal idea expounded in [Qui63, Har86] among others is made precise in the succeeding chapters so that we can design a machine that obeys this principle. Chapter 3 The Theory of Irrelevance 3.1 Introduction Often, wisdom is knowing what to ignore. An autonomous resource limited agent with a very detailed theory of the world should ....

....akin to Snell s law for computational systems: they propose taking the path of least resistance in the space of representations and computations. The representational irrelevance principle is a variation of Quine s principle of indiscernibility of identicals [Qui63] It is re echoed in Harman s [Har86] principle of clutter avoidance and Lenat s notion of cognitive economy [LHRK79] Whereas the representational irrelevance principle outlaws making unnecessary distinctions, the computational irrelevance principle outlaws unneeded computation. To apply the computational irrelevance principle we ....

G. Harman. Change in View. MIT Press, Cambridge, MA, 1986.

.... comprehender has available both weak schemas, which locally link small amounts of input (e.g. causal schemas) and strong schemas, which globally link larger sections of input (e.g. scripts) All schemas generate connections of intelligibility which affect the coherence of a representation (Harman, 1986). Coherence is a common currency with which to measure the benefit of applying a schema. Instead of requiring that a top level structure be instantiated, the system instead applies schemas to produce a representation of sufficient value . This process can be naturally described as abduction, or ....

G. Harman. 1986. Change in View. MIT Press, Cambridge, MA.

....resulting theorem prover, OSCAR, would be quite slow in comparison to theorem provers designed to take advantage of the special abilities of computers. Instead, OSCAR has turned out to be surprisingly efficient, doing some hard problems faster than most resolution1 1 For instance, Lusk et al. [9], 10] Stickel [19] 20] Wos and Winker [21] and others too numerous to mention. based theorem provers. The deductive reasoner described here is part of a more general reasoner that performs both deductive and defeasible reasoning. The original reason for exploring non resolution reasoners ....

....from the perspective of the epistemologist, who is primarily interested in understanding the structure of human rational thought. Human rational architecture is described by rules for updating one s set of beliefs. A feature of these rules that has recently been emphasized by Gilbert Harman [9] is that they do not mandate the adoption of every belief for which we have good reasons. For example, each belief P gives us a conclusive reason for any disjunction (P Q) containing P as a disjunct, but we do not arbitrarily adopt such disjunctions. Our epistemic rules must honor the fact that ....

Harman, G., Change in View (MIT press, 1986).

....equally important to consider the question of belief revision, that is, the situations which may cause a change of beliefs and the conditions under which particular changes occur. The topic of belief revision is in fact a very active research area in both AI (e.g. Vardi, 1988) and philosophy (e.g. Harman, 1986). Harman, for example, offers some informal principles of belief revision, such as, that one should make minimal changes to one s view that increase its coherence as much as possible while promising suitable satisfaction of one s ends . Which beliefs one discards depends on the justifications one ....

Harman, G. (1986). Change in View. Cambridge, Mass.: MIT Press.

.... IRS, for the following reasons: ffl As in the case of fuzziness, how the degree of belief is related to weight of evidence is unclear ( 21] ffl Due to insufficient knowledge and resources, it is usually impossible for IRS to maintain a consistent probability assignment on its knowledge base ([9]) ffl As discussed in [28] conditionalization cannot be referred as a general way to symmetrically combine evidence from different sources. ffl With all the efforts to improve its efficiency, the resources expense of Bayesian approach is still pretty high for large knowledge bases. The ....

G. Harman. Change in View. The MIT Press, Cambridge, Massachusetts, 1986.

No context found.

Harman, G. (1986). Change in View. Cambridge, Mass.: mit Press.

No context found.

Gilbert Harman. Change in View. MIT Press, Cambridge, MA, 1986.

No context found.

Gilbert Harman. 1986. Change of View. MIT PRESS.

No context found.

Harman, G. (1986) Change In View, Bradford Books, MIT Press, Cambridge, MA.

No context found.

G. Harman. Change in View. MIT Press, Cambridge, MA, 1986.

First 50 documents

Online articles have much greater impact   More about CiteSeer.IST   Add search form to your site   Submit documents   Feedback  

CiteSeer.IST - Copyright Penn State and NEC