BfoCitationsDistinguishingDispositionFunctionAndRole

BFO:Realizable Entity: Distinction between Disposition, Function, and Role: BFO Citations providing distinguishing definitions/axioms for the children of bfo:realizable_entity

This discussion sub-thread took place on the BFO mailing list in July 2007. It is summarized here, since it might provide us with a clearer sense of how to proceed with OBI:Function, which must build off one or more of these BFO realizable entity types. The discussion was heavily edited to hopefully promote clarity - and referencing URLs have been added to help point others to critical background material.
The thread needed to be broken into sub-threads, because it extended beyond the 32kb MediaWiki recommended page limit.

<< Back to main page for this discussion
<< Back to OBI Function BFO + Use Case page
<< Back to OBI Function maing page
<< Back to Homepage

Discussion Participants

• Bill Bug (BB)
• Cristian Coscos (CC)
• Jennifer Fostel (JF)
• Pierre Grenon (PG)
• Robert Hoehndorf (RH)
• Waclaw Kusnierczyk (WK)
• Kristl Laux (KL)
• Matthias Samwald (MS)
• Barry Smith (BS)
• Alan Ruttenberg (AR)

BFO Citations providing distinguishing definitions/axioms for the children of bfo:realizable_entity

• BS - July 18, 2007 3:49:23 AM EDT
  • See this earlier BFO list discussion on disposition', function, and role (Summary follows)
    • AR - March 1, 2007
      • Is there a paper with more formal defining axioms for the subtypes of realizable_entity more so than the cornucopia paper? I'm still trying to clarify the distinctions between role, disposition, function, power, etc.
      • PG - March 1, 2007 06:31 AM EDT
        • There is nothing to my knowledge. It is possible however that some elements have been given in work on functions, maybe papers authored or co-authored by Ingvar Johansson. I will ask him. Maybe Barry will have better suggestions.
          • BS - March 3, 2007 03:31 PM EDT
            • There is nothing, as yet, except for Ingvar's and Ludger's work on tendencies:
              1. The Ontology of Tendencies and Medical Information Sciences
              2. INTENTIONALITY AND TENDENCY: HOW TO MAKE ARISTOTLE UP-TO-DATE
            • We have been holding back until other matters solidify, but if you have proposals I would be eager to hear them.
            • First ideas:
              1. function
                 • definition: disposition that is beneficial to the bearer
              2. role
                 • definition: socially assigned function (or more likely: constellation of functions)
              3. power
                 • definition: disposition whose realization gives rise to the release of energy
              4. liability
                 • definition: disposition whose realization involves the absorption of energy
              5. tendency
                 • definition: ?
              6. propensity
                 • definition: disposition with a statistically determinate likelihood of being realized
            • BFO probably should not include all of these
            • Counterexamples welcome.
              • RH - March 7, 2007 06:32 AM EDT
                • Re: function
                  • I have a disposition to eat. Eating is beneficial for me. My function is to eat. Seems counter-intuitive to me. The problem is that beneficial depends on values of some observer, something you probably do not want. The basic idea of the realist accounts on saying what beneficial means is:
                    • A can cause X. A is there now because it caused X in the past.
                  • I believe you would like to reduce function to purely realistic terms without human ascription. The best work I know of that tries to do this is Ruth Millikans Language, Thought and other biological categories.
                  • See Searle for a number of additional counter-examples.
      • PG - March 1, 2007 06:31 AM EDT
        • To my knowledge, we have no worked on any of the specificities of each sort of realizable_entity - I.e. we have not worked on the specificity of each sort of realization (processes and relations).
    • AR - March 1, 2007
      • As I see it:
        • "Realization involves in every case three entities:
          1. a substance
          2. a SNAP:dependent_entity
          3. a correlated process
        • (2) then inheres_in (1) in virtue of the fact (1) participates in (3)" (however correlated is a bit fuzzy)
          • BS - March 3, 2007 03:31 PM EDT
            • A disposition would then involve a correlated process (3) universal. The realization of the disposition would be an instance.
    • WK - July 18, 2007 4:28:54 AM EDT
      • Clearly, Alan expressed his doubts four months ago, and if he asks again, he must have not received an appropriate explanation.
      • Pierre seems to suggest role, function, power, liability, propensity are all types of disposition; this fits Matthias' view.
        • CC - July 19, 2007 4:27:46 AM EDT
          • Pierre's right, as always. I'd even go further and suggest making a distinction between power, liability, propensity, tendency or whatnot is an overly fastidious enterprise, bordering on squeamishness. I do not think such distinctions are consecrated enough as to be reflected in an ontology, much less in an upper-level one.
            • WK - July 19, 2007 4:53:04 AM EDT: The goal of BFO is to represent reality as it is, not as it might be useful to describe it, or as our cognition may wish to see it. Ask Pierre what there is, and represent what he says there is. Done.
              • BB - July 24, 2007 11:30 AM EDT:
                • I can understand WKs frustration. Having said that, I don't think we seek advice from Pierre, Cristian, Barry, Ingvar, Fabian, Holger, etc., because they are the keepers of the law. It is because they are - as Waclaw himself described them earlier in this thread - trained, practicing (realist) philosophers with extensive application experience who have been associated directly or indirectly with the development of BFO. We need the benefit of their knowledge and experience in that field, as we work to put BFO to effective use in biomedical informatics.
                • The problem - and the origin of the tension we are wrestling with in this discussion - relates to the fact a large collection of complex biomedical ontologies are being built which we all see as critical to the next stage of biomedical informatics development in the 21st century. We need these ontologies now - yesterday, really - and it appears we're relying on a foundational ontology where some critical issues are still under active debate. I very much appreciate the time being invested by all in working out these issues.
                • I believe all would also agree this feedback from us relatively knowledgeable users is a very important part of that process.
                • Having said that - even while the debate is ongoing - I don't see why we can't immediately improve the definitions and examples in the current BFO OWL file NOW, so we can keep moving the ontologies built on BFO forward - especially given these OWL AnnotationProperties will have no effect on how the DIG reasoners process the ontology and but would significantly improve our ability to put BFO to correct and effective use. I would also include implementing some of the examples in OWL using the current BFO OWL files (v1.0 AND v1.1). If in the course of improving the definitions and examples, we resolve some of these outstanding issues, all the better.
          • The moral is BFO in its present version covers all such categories.
          • Turning function into a child of disposition would make perfect sense in my opinion.
            • WK - July 19, 2007 4:53:04 AM EDT: Now we need to hear from the BFO's custodians
              1. your suggestion will be implemented, or
              2. a counterargument to what Matthias has proposed.
          • Note: I have stopped asking for the demise of the disposition category as, I'm being told, that would certainly anger classically-trained biologists.
            • Question: Should BFO bow before creatures (like me) who have no idea about philosophy and do not really know the existing work in that field, and yet demand (not like me) BFO include such useful universals as disposition? (WK - July 19, 2007 4:53:04 AM EDT)
      • As Robert remarks, beneficial and socially assigned are not particularly precise expressions.
      • Furthermore, I see there would be multiple inheritance: there can be dispositions both beneficial to the bearer, socially assigned, and giving rise to the release of energy and have a statistically determinate likelihood of being realized -- right? Aren't those categories (role, function, etc.) supposed to be disjoint? Or in cases disposition fits into each set, would one have a separate instance of each category -- to keep them disjoint? e.g., a natural transplanted heart would have the function to pump blood and, separately, the role to pump blood, and the power to pump blood, and a propensity to pump blood...?