Directories | Web | Images | Groups | News | Shopping | Local

Enter your search keyword(s):

 



(formerly Encyclopedic directory)
Belief Networks
Home / Top / Computers / Artificial Intelligence / Belief Networks See also:
Related articles

Edit | Discuss Article

Belief

Belief is assent to a proposition.

Belief in the psychological sense, is a representational mental state that takes the form of a propositional attitude. In the religious sense, "belief" refers to a part of a wider spiritual or moral foundation, generally called faith.

Belief is considered propositional in that it is an assertion, claim or expectation about reality that is presumed to be either true or false (even if this cannot be practically determined, such as a belief in the existence of a particular deity).

Historically, philosophical attempts to analyze the nature of belief have been couched in terms of judgement. Both David Hume and Immanuel Kant are both particularly well known for their analyses using this framework.

Table of contents
1 Belief, knowledge and epistemology
2 Belief as a psychological theory
3 Is belief voluntary?
4 Delusional beliefs
5 See also

Belief, knowledge and epistemology

Knowledge is often defined as justified true belief, in that the belief must be considered to correspond to reality and must be derived from valid evidence and arguments. However, this definition has been challenged by the Gettier problem which suggests that justified true belief does not provide a complete picture of knowledge.

To believe something can be interpreted as assigning a probability of more than 50% that something is true. The rule of the thumb from a school of epistemology that says that certainty should be as big as the corresponding evidence is called evidentialism.

Belief as a psychological theory

Mainstream psychology and related disciplines have traditionally treated belief as if it were the simplest form of mental representation and therefore one of the building blocks of conscious thought. Philosophers have tended to be more rigorous in their analysis and much of the work examining the viability of the belief concept stems from philosophical analysis.

The concept belief presumes a subject (the believer) and an object of belief (the proposition) so like other propositional attitudes, belief implies the existence of mental states and intentionality, both of which are hotly debated topics in the philosophy of mind and whose foundations and relation to brain states are still controversial.

Beliefs are sometimes divided into core beliefs (those which you may be actively thinking about) and dispositional beliefs (those which you may ascribe to but have never previously thought about). For example, if asked 'do you believe tigers wear pink pyjamas ?' a person might answer that they do not, despite the fact they may never have thought about this situation before.

The idea that a belief is a mental state is much more contentious. While some philosophers have argued that beliefs are represented in the mind as sentence-like constructs others have gone as far as arguing that there is no consistent or coherent mental representation that underlies our common use of the belief concept and is therefore obsolete and should be rejected.

This has important implications for understanding the neuropsychology and neuroscience of belief. If the concept of belief is incoherent or ultimately indefensible then any attempt to find the underlying neural processes which support it will fail. If the concept of belief does turn out to be useful then this goal should (in principle) be achievable.

Philosopher Lynne Rudder Baker has outlined four main contemporary approaches to belief in her book Saving Belief:

  • Our common-sense understanding of belief is correct - Sometimes called the ‘mental sentence theory’, in this conception, beliefs exist as coherent entities and the way we talk about them in everyday life is a valid basis for scientific endeavour. Jerry Fodor is one of the principle defenders of this point of view.
  • Our common-sense understanding of belief may not be entirely correct, but it is close enough to make some useful predictions - This view argues that we will eventually reject the idea of belief as we have use it now, but that there may be a correlation between what we take to be a belief when someone says 'I believe that snow is white' and however a future theory of psychology will explain this behaviour. Most notably philosopher Stephen Stich has argued for this particular understanding of belief.
  • Our common-sense understanding of belief is entirely wrong and will be completely superseded by a radically different theory will have no use for the concept of belief as we know it - Known as eliminativism, this view, most notably proposed by Paul and Patricia Churchland), argues that the concept of belief is like obsolete theories of times past such as the four humours theory of medicine, or the phlogiston theory of combustion. In these cases science hasn’t provided us with a more detailed account of these theories, but completely rejected them as valid scientific concepts to be replaced by entirely different accounts. The Churchlands argue that our common-sense concept of belief is similar, in that as we discover more about neuroscience and the brain, the inevitable conclusion will be to reject the belief hypothesis in its entirety.
  • Our common-sense understanding of belief is entirely wrong, however treating people, animals and even computers as if they had beliefs, is often a successful strategy - The major proponents of this view, Daniel Dennett and Lynne Rudder Baker, are both eliminativists in that they believe that beliefs are not a scientifically valid concept, but they don’t go as far as rejecting the concept of belief as a predictive device. Baker gives the example of playing a computer at chess. While few people would agree that the computer held beliefs, treating the computer as if it did (e.g. that the computer believes that taking the opposition’s queen will give it a considerable advantage) is likely to be a successful and predictive strategy. In this understanding of belief, named by Dennett the intentional stance, belief based explanations of mind and behaviour are at a different level of explanation and are not reducible to those based on fundamental neuroscience although both may be explanatory at their own level.

Is belief voluntary?

Most philosophers hold the view that belief formation is to some extent spontaneous and involuntary. Some people think that one can choose to investigate and research a matter but that one can not choose to believe. On the other hand, most people have the impression that in some cases people don't believe things because they don't want to believe, especially about a matter in which they are emotionally involved.

Delusional beliefs

Delusions are defined as beliefs in psychiatric diagnostic criteria (for example in the DSM). Psychiatrist and historian German Berrios has challenged the view that delusions are genuine beliefs and instead labels them as "empty speech acts", where affected persons are motivated to express false or bizarre belief statements due to an underlying psychological disturbance. However, the majority of mental health professionals and researchers treat delusions as if they were genuine beliefs.

See also

 

Source | Copyright
Webmasters: Add your website here:

Readers: Edit | Discuss Listings

A Brief Introduction to Graphical Models and Bayesian Networks
Kevin Murphy's tutorial, including a recommended reading list.
http://www.cs.berkeley.edu/~murphyk/Bayes/bayes.html

B-Course - Dependence and classification modeling
A free, interactive tutorial on Bayesian modeling, in particular dependence and classification modeling.
http://b-course.cs.helsinki.fi

Cause, chance and Bayesian statistics
Briefing document with a short survey of Bayesian statistics
http://www.abelard.org/briefings/bayes.htm

Learning Bayesian Networks from Data
Slides and additional notes from a tutorial by Nir Friedman and Daphne Koller on automated learning of belief networks, given at the Neural Information Processing Systems (NIPS-2001) conference
http://www.cs.huji.ac.il/~nirf/Nips01-Tutorial/

Belief Revision
Software, publications, teaching material, and news on belief revision - from the Business and Technology Research Laboratory at the University of Newcastle, Australia
http://beliefrevision.org

Daphne's Approximate Group of Students (DAGS)
Daphne Koller's research group on probabilistic representation, reasoning, and learning at Stanford University
http://dags.stanford.edu

Bayesian Network Repository
Maintained by Gal Elidan - over a dozen publicly available networks with documentation, in several popular interchange formats
http://www.cs.huji.ac.il/labs/compbio/Repository/

An Introduction to Bayesian Networks and Their Contemporary Applications
A survey and tutorial by Daryle Niedermayer - covers material on Bayesian inference in general and selected industrial applications of graphical models
http://www.niedermayer.ca/papers/bayesian/

Decision Systems Lab (DSL)
Research group at the University of Pittsburgh with links to books and software on probabilistic, decision-theoretic, and econometric graphical models
http://www.sis.pitt.edu/~dsl/

LAPLACE Group - Bayesian Models for Perception, Inference and Action
Probabilistic reasoning and genetic algorithms for perception, inference and action: Bayesian cognitive and brain models, software for robotics, probabilistic inference engine
http://www-laplace.imag.fr

Association for Uncertainty in Artificial Intelligence
Main association for belief network researchers. Runs the annual Uncertainty in Artificial Intelligence (UAI) conferences, and the UAI mailing list.
http://www.auai.org/

Qualitative Verbal Explanations in Bayesian Belief Networks
Paper about combining probabilistic models and human-intuitive approaches to modeling uncertainty by generating qualitative verbal explanations of reasoning.
http://www.pitt.edu/~druzdzel/abstracts/aisb.html

Query DAGs: A Practical Paradigm for Implementing Belief-Network Inference
Article published in JAIR (Journal of AI Research) about a way to implement belief networks by compiling networks into arithmetic expressions and then answering queries using an evaluation algorithm.
http://www.cs.cmu.edu/afs/cs/project/jair/pub/volume6/darwiche97a-html/jair-f.html

Belief Networks and Variational Methods : Amos Storkey
Dynamic Trees are mixtures of tree structured belief networks, and are used as models for image segmentation and tracking.
http://www.anc.ed.ac.uk/~amos/belief.html

UConn list of Bayesian Network Resources
Eugene Santos' lists of belief network research, papers, and systems.
http://excalibur.brc.uconn.edu/~baynet/



Help build the largest human-edited directory on the web.
 Submit a Site - Open Directory Project (modified) - Become an Editor

Modified contents copyright 2008. All rights reserved.