Definitions (continuant, independent continuant)

154 views
Skip to first unread message

Piotr Nowara

unread,
Feb 10, 2010, 4:39:35 PM2/10/10
to BFO Discuss
Dear all,

I'd like to share some thoughts on the definitions of continuant and
independent continuant:

1a) Continuant is defined as “entity [bfo:Entity] that exists in full
at any time in which it exists at all, persists through time while
maintaining its identity and has no temporal parts”.

What does it mean „to exist in full”? How „not-existing in full” would
differ from „existing in full”? Should I say that „I do exist in full
during this very moment of my life, but the process of my life exists
in 30%”? What happens if the process of my life exists in 100%? What
is the meaning of these kind of statements? How this quantification
of existing could be of any use?

In my opinion this is an example of confusing existence with
completeness. In fact the term „completeness” should be applied only
to some occurents, more precisely to processes (maybe it could be also
useful to describe events as something that is completed int the
moment of occuring).

1b) “(...) persists through time while maintaining its identity”
does it mean that all continuants are entities that persist through
time?

2) Definition of independent continuant doesn't exclude non-material
objects, however there is an equivalent class restriction on the
IndependentContinuant class that makes BFO a materialistic ontology:
EquivalentClasses(snap:IndependentContinuant
ObjectUnionOf(snap:MaterialEntity snap:ObjectBoundary snap:Site)).

What is the purpose of this restriction? Why it adds so much to the
definitions presented in the annotations?

Piotr Nowara

Phillip Lord

unread,
Feb 11, 2010, 6:42:20 AM2/11/10
to bfo-d...@googlegroups.com

Piotr Nowara <piotr...@gmail.com> writes:
> I'd like to share some thoughts on the definitions of continuant and
> independent continuant:
>
> 1a) Continuant is defined as “entity [bfo:Entity] that exists in full
> at any time in which it exists at all, persists through time while
> maintaining its identity and has no temporal parts”.
>
> What does it mean „to exist in full”? How „not-existing in full” would
> differ from „existing in full”? Should I say that „I do exist in full
> during this very moment of my life, but the process of my life exists
> in 30%”? What happens if the process of my life exists in 100%? What
> is the meaning of these kind of statements? How this quantification
> of existing could be of any use?


I understand why you find the definitions confusing. It's a lot easier
to understand though if you just ignore the philosophy and think of a
pragmatic application.

If we think about me (my favourite topic of conversation). I find it
useful to think of myself as being the same person as I was in the past,
even though most of the bits of me have been mostly replaced compared to
me when I was 6 months old.

On the other hand, my life is not yet finished, and hopefully won't be
by the time you read this. So, my life between my birth (conception,
ensoulment, whatever) and now is not all of my life.


> In my opinion this is an example of confusing existence with
> completeness. In fact the term „completeness” should be applied only
> to some occurents, more precisely to processes (maybe it could be also
> useful to describe events as something that is completed int the
> moment of occuring).
>
> 1b) “(...) persists through time while maintaining its identity”
> does it mean that all continuants are entities that persist through
> time?

It means that a continuant while it maintains its identity, persists.
You can't have a gap. So, again using me as a example, I am here now,
and I will persist until I die. After that I won't persist and never
again will my like tread this earth. That is once I die, there will be
no object that exists which is identical to me; also, before I was born,
there was no object identical to me.

> 2) Definition of independent continuant doesn't exclude non-material
> objects, however there is an equivalent class restriction on the
> IndependentContinuant class that makes BFO a materialistic ontology:
> EquivalentClasses(snap:IndependentContinuant
> ObjectUnionOf(snap:MaterialEntity snap:ObjectBoundary snap:Site)).
>
> What is the purpose of this restriction? Why it adds so much to the
> definitions presented in the annotations?


I can't comment on that on the value of this. BFOs idea that only things
with mass are terribly important is broken and is one of the things that
leads to the conclusion that a spatial region can't have a length.

Phil

Wacek Kusnierczyk

unread,
Feb 11, 2010, 7:02:22 AM2/11/10
to bfo-d...@googlegroups.com
Phillip Lord wrote:
> Piotr Nowara <piotr...@gmail.com> writes:
>
>
>> 1b) �(...) persists through time while maintaining its identity�

>> does it mean that all continuants are entities that persist through
>> time?
>>
>
> It means that a continuant while it maintains its identity, persists.
> You can't have a gap. So, again using me as a example, I am here now,
>

so let's use another example. maybe the ship of theseus? or maybe a
watch: it's here, i dismount it, there are only the parts, does the
watch persist (in full, that is)? if i mount the parts back again, in
the same configuration, is it *the* watch, or another one? if the
former, did it persist with no gap? how do you tell the *identity* in
this case?

if, in the meantime, the parts are, temporarily, parts of some other
watches, does *the* watch still persist when it's not there? and if i
decide not to mount the parts back, when does the watch cease to
persist: when it was first dismounted, when i decided not to mount it,
when any of it's parts ceases to exist (and that seems to call for a
recursive question), or never?

vQ

Pierre Grenon

unread,
Feb 11, 2010, 7:38:25 AM2/11/10
to bfo-d...@googlegroups.com
Hi Piotr,

On Wed, Feb 10, 2010 at 9:39 PM, Piotr Nowara <piotr...@gmail.com> wrote:

> 2) Definition of independent continuant doesn't exclude non-material
> objects, however there is an equivalent class restriction on the
> IndependentContinuant class that makes BFO a materialistic ontology:
> EquivalentClasses(snap:IndependentContinuant
> ObjectUnionOf(snap:MaterialEntity snap:ObjectBoundary snap:Site)).
>
> What is the purpose of this restriction? Why it adds so much to the
> definitions presented in the annotations?

This is a mistake and the axiom should go away. Thank you for pointing this.

Independently of the merits of introducing MaterialEntity on which I
will not comment, the introduction of MaterialEntity was supposed to
be done with the understanding that it would not preclude the
possibility of immaterial continuants. It is also enough that a
materialist version of BFO can be created to satisfy contextual needs
by adding this axiom.

Incidentally, I don't know that I agree with the disjointness axioms
on MaterialEntity. This should probably remain open at this level but
maybe it is arguable.

best
pierre

> Piotr Nowara

Pierre Grenon

unread,
Feb 11, 2010, 7:41:13 AM2/11/10
to bfo-d...@googlegroups.com
> BFOs idea that only things

I question the 'only'

> with mass are terribly important is broken and is one of the things that
> leads to the conclusion that a spatial region can't have a length.

didn't we go over this? Nothing prevents you from ascribing a length
to a region. But yes, BFO doesn't claim that there is an entity which
is the length of a region.

cheers
p

> Phil

Pierre Grenon

unread,
Feb 11, 2010, 8:24:12 AM2/11/10
to bfo-d...@googlegroups.com
There should be discussions of these in the list's archive, but I'll
give a try to answering your questions.

On Wed, Feb 10, 2010 at 9:39 PM, Piotr Nowara <piotr...@gmail.com> wrote:

> Dear all,
>
> I'd like to share some thoughts on the definitions of continuant and
> independent continuant:
>
> 1a) Continuant is defined as “entity [bfo:Entity] that exists in full
> at any time in which it exists at all, persists through time while
> maintaining its identity and has no temporal parts”.
>
> What does it mean „to exist in full”?

The phrase that is intended to be meaningful is "to exist in full at a
given time". But even this sentence is possibly as clear as mud.

Does it help to say that when you're considering an entity of the
intended sort at a given time, you don't go looking for it at other
times? Perhaps it's useful to think of it in the following way: for
entities that exist at more than one time, it boils down to not having
temporal parts. But this is little more than a gloss on a formal
trick.

The caveat is that this is useful as long as you understand it as
typical of the persisting members of the kind. It is no longer a
useful distinction if you start contemplating instantaneous members.
To come to term with the fact that there could be instantaneous
members in the relevant kinds and that the kinds remain nevertheless
disjoint, you have to first accept the distinction for persisting
entities, then consider a particular kind, and imagine what an
instantaneous members of the special kind would be like.

For example, accept that there is a distinction between a car and its
life. Then imagine an instantaneous car. It still is distinct from its
life, under the framework, even if both are instantaneous.

> How „not-existing in full” would
> differ from „existing in full”?

The first entity would have temporal parts existing at other times.

> Should I say that „I do exist in full
> during this very moment of my life, but the process of my life exists
> in 30%”?

You could say this, if you live 3.33 moments.

> What happens if the process of my life exists in 100%?

It does over your lifetime, the point is that your lifetime is
extended. The other case is when your life is instantaneous, see
above.

> What
> is the meaning of these kind of statements?

Just don't try to build too much in them.

They are just attempts at illustrating a fundamental distinction which
is hard to illustrate and they work only until to some extent. They
are useful under the assumption that the entities you are talking
about are timespanning entities and exist for more than an instant.
They do not allow you to distinguish between instantaneous entities
that fall in the two different main kinds of entities in BFO.

> How  this quantification
> of existing could be of any use?

People tend to request definitions for things which are either hard to
define or ought to not be defined at all.

What matters is the fact that there are two distinct kinds. And then,
what is important in relation to the distinction is the formal theory
that accompanies the kinds of entity delineated and the way they
relate. This is not in the OWL file.

> In my opinion this is an example of confusing existence with
> completeness. In fact the term „completeness” should be applied only
> to some occurents, more precisely to processes (maybe it could be also
> useful to describe events as something that is completed int the
> moment of occuring).

No, it's two kinds of entity. Perhaps they are poorly characterised by
a brief sentence drawing attention to distinct ways in which
persisting members persist. The description may be suboptimal but it
should not lead to making what it describes something that fits some
understanding or another of a peculiar phrase.

> 1b) “(...) persists through time while maintaining its identity”
> does it mean that all continuants are entities that persist through
> time?

BFO remains open, I think, and accepts the possibility of
instantaneous continuants, although it doesn't particularly rely on
it.

cheers
p

>
> Piotr Nowara

Phillip Lord

unread,
Feb 11, 2010, 8:32:25 AM2/11/10
to bfo-d...@googlegroups.com
Wacek Kusnierczyk <wa...@idi.ntnu.no> writes:
> Phillip Lord wrote:
>> Piotr Nowara <piotr...@gmail.com> writes:
>>
>>
>>> 1b) “(...) persists through time while maintaining its identity”

>>> does it mean that all continuants are entities that persist through
>>> time?
>>>
>>
>> It means that a continuant while it maintains its identity, persists.
>> You can't have a gap. So, again using me as a example, I am here now,
>>
>
> so let's use another example. maybe the ship of theseus? or maybe a watch:
> it's here, i dismount it, there are only the parts, does the watch persist (in
> full, that is)? if i mount the parts back again, in the same configuration,
> is it *the* watch, or another one? if the former, did it persist with no gap?
> how do you tell the *identity* in this case?

Well, I am far from being an apologist for BFO, as you know, so I don't
necessarily have answers to this. But this is not a new issue:

"This is the original axe that George Washington used; it's head three
new handles and two new heads since then, but it's the same axe".

> if, in the meantime, the parts are, temporarily, parts of some other watches,
> does *the* watch still persist when it's not there? and if i decide not to
> mount the parts back, when does the watch cease to persist: when it was first
> dismounted, when i decided not to mount it, when any of it's parts ceases to
> exist (and that seems to call for a recursive question), or never?

I also do not have an answer to this; rather, I don't think that there
is a single comprehensive answer which works in all circumstances. We
need to make a judgement about what we want from the notion of identity,
based upon a particular set of use cases.

So, for example, in most cases this would be considered to not be the
same watch; certainly, it would not be if you tried to use your
guarantee. The point here is that in a commodity engineering
environment, it is not intended that you take your watch to bits and put
it back together again.

In a scientific or specialist context, however, this is not true.
Aircraft engines, for example, are regularly disassembled and rebuilt as
part of their service schedule. If we have to describe them as new
engine every time it's going to make the task of handling any data about
them really very hard.

A design pattern like BFO can be useful, but this doesn't avoid the
necessity for clear use cases. What is the best answer to your questions
depends on what you want to do.

This would be my take on it anyway.

Phil

Barry Smith

unread,
Feb 11, 2010, 8:49:58 AM2/11/10
to bfo-d...@googlegroups.com
I agree with Phil.
There are in fact two (old) problems here: the watch repair problem
(intermittent existence, multiple beginnings of existence), and the
Theseus problem (vagueness of identity and of time when artifacts
begin or cease to exist). People have attempted to resolve these
problems, now, for two millennia, and will probably continue to fail
in these attempts. Any alternative to BFO will be marked by the same
problems, or by analogous problems at some other place. But they are,
experience suggests, not fatal. For they do not affect the vast
majority of cases to which BFO is applied, and even where they do
arise (for example engine repair) strategies can be adopted to fit.
BS

>--
>You received this message because you are subscribed to the Google
>Groups "BFO Discuss" group.
>To post to this group, send email to bfo-d...@googlegroups.com.
>To unsubscribe from this group, send email to
>bfo-discuss...@googlegroups.com.
>For more options, visit this group at
>http://groups.google.com/group/bfo-discuss?hl=en.

Wacek Kusnierczyk

unread,
Feb 11, 2010, 9:08:52 AM2/11/10
to bfo-d...@googlegroups.com
Phillip Lord wrote:
> Wacek Kusnierczyk <wa...@idi.ntnu.no> writes:
>
>>
>> so let's use another example. maybe the ship of theseus? or maybe a watch:
>> it's here, i dismount it, there are only the parts, does the watch persist (in
>> full, that is)? if i mount the parts back again, in the same configuration,
>> is it *the* watch, or another one? if the former, did it persist with no gap?
>> how do you tell the *identity* in this case?
>>
>
> Well, I am far from being an apologist for BFO, as you know, so I don't
> necessarily have answers to this. But this is not a new issue:
>

of course not, that's why i referred to the ship of theseus.

> "This is the original axe that George Washington used; it's head three
> new handles and two new heads since then, but it's the same axe".
>
>
>> if, in the meantime, the parts are, temporarily, parts of some other watches,
>> does *the* watch still persist when it's not there? and if i decide not to
>> mount the parts back, when does the watch cease to persist: when it was first
>> dismounted, when i decided not to mount it, when any of it's parts ceases to
>> exist (and that seems to call for a recursive question), or never?
>>
>
> I also do not have an answer to this; rather, I don't think that there
> is a single comprehensive answer which works in all circumstances. We
> need to make a judgement about what we want from the notion of identity,
> based upon a particular set of use cases.
>

possibly, but i don't think bfo would be willing to adapt to your (or
mine) particular cases, as it purports to be a very general and
all-applicable theory.

> So, for example, in most cases this would be considered to not be the
> same watch; certainly, it would not be if you tried to use your
> guarantee.

actually, i think it would, provided that there would be somewhere there
an id number, as stated in some guarantee document. it might not be
considered for unpaid repair, though, if the disassembly were obvious.


> The point here is that in a commodity engineering
> environment, it is not intended that you take your watch to bits and put
> it back together again.
>

well, yes, though one could always find examples such as an mp3 set,
where you regularly detach the earphones, maybe use them with different
devices, and ask whether afterwards it's the same set. whatever, it
doesn't matter. it's about principles.

> In a scientific or specialist context, however, this is not true.
> Aircraft engines, for example, are regularly disassembled and rebuilt as
> part of their service schedule. If we have to describe them as new
> engine every time it's going to make the task of handling any data about
> them really very hard.
>

not necessarily, it's just about an appropriate model to that. the
question is what you assign an id to. quite similar to issues in
genomic databases.

> A design pattern like BFO can be useful, but this doesn't avoid the
> necessity for clear use cases. What is the best answer to your questions
> depends on what you want to do.
>
> This would be my take on it anyway.
>

i agree; this leaves two options: 1) produce a version of bfo
appropriate to the situations where different patterns would be needed;
2) don't use bfo where the particular, fixed choice is not appropriate
for the specific need.

or?

vQ

Phillip Lord

unread,
Feb 11, 2010, 10:17:40 AM2/11/10
to bfo-d...@googlegroups.com


Yes, we did go over this. I am correct in my assertion. Length is a
quality. Qualities inhere only in IndependentContinuant. A spatial
region is not one.

Phil

Phillip Lord

unread,
Feb 11, 2010, 10:41:40 AM2/11/10
to bfo-d...@googlegroups.com
Wacek Kusnierczyk <wa...@idi.ntnu.no> writes:
>> I also do not have an answer to this; rather, I don't think that there
>> is a single comprehensive answer which works in all circumstances. We
>> need to make a judgement about what we want from the notion of identity,
>> based upon a particular set of use cases.
>
> possibly, but i don't think bfo would be willing to adapt to your (or mine)
> particular cases, as it purports to be a very general and all-applicable
> theory.

Then BFO would be wrong.

However, I think you can look at it in another way. Just because we
don't have a good way of understanding identify for all circumstances,
does not mean that we don't need the concept of identity in many.

To use a programming example (and I realise that these often break), we
may want all objects to be comparable for identity, but this doesn't
mean that all types of objects implement this identify check in the same
way.

>> So, for example, in most cases this would be considered to not be the
>> same watch; certainly, it would not be if you tried to use your
>> guarantee.
>
> actually, i think it would, provided that there would be somewhere there an id
> number, as stated in some guarantee document. it might not be considered for
> unpaid repair, though, if the disassembly were obvious.

Okay:-)

>> In a scientific or specialist context, however, this is not true.
>> Aircraft engines, for example, are regularly disassembled and rebuilt as
>> part of their service schedule. If we have to describe them as new
>> engine every time it's going to make the task of handling any data about
>> them really very hard.
>
> not necessarily, it's just about an appropriate model to that. the question
> is what you assign an id to. quite similar to issues in genomic databases.

Sure. What I am saying in this case, you want to apply identity despite
the disassembly, as this would be the easiest implementation path. If I
were designing a database, I'd probably have an id for the engine, and a
build number. Data like flight time, for example, applies to the engine
since it was first constructed.

>> A design pattern like BFO can be useful, but this doesn't avoid the
>> necessity for clear use cases. What is the best answer to your questions
>> depends on what you want to do.
>>
>> This would be my take on it anyway.
>
> i agree; this leaves two options: 1) produce a version of bfo appropriate to
> the situations where different patterns would be needed; 2) don't use bfo
> where the particular, fixed choice is not appropriate for the specific need.

Or, 3) accept that ontologies which are built with different intentions
in mind are not going to be trivially integrated, even if they are based
on BFO. To my mind, you are going to get 3) whether you want it or not.

Let me explain: one solution to the problem is this. You could define 5
or 10 different kinds of identity -- following on from these examples
"identity preserved over disassembly" and "identity destroyed by
disassembly". Now this would work. The reassembled watch would be a new
object, and reassembled jet engine would not be, and there would be no
integration problems because we have a different form of identity so we
would expect it to be different.

However, if you really, really want to integrate your jet engine and
watch ontology, that's not good enough; it's good that you've been
explicit and you know up front that it's not going to integrate, but it
doesn't help you to integrate.

Worse, these two identities are not going to be enough; I, for example,
have never had a leg replaced, but all the atoms change regularly. So,
that's three forms of identity. Eventually, once we get to 5 or 10, it's
going to get so complex, that no mortal will be able to work it out
anyway; so the forms of identity will be applied inconsistently. Again,
once you come to integrate ontologies, you will have to do lots of work
because of this inconsistency.

There are a number of upper ontologies around. They help somewhat, they
can get you going. But, they are not going to allow automatic
integration because different people will make different, but
appropriate decisions, at different times.

Phil

Barry Smith

unread,
Feb 11, 2010, 10:44:20 AM2/11/10
to bfo-d...@googlegroups.com

It is clear that BFO needs a strategy to allow predications of this
sort even where there are no BFO:qualities (thus we need also to
allow predications where the subject of the predication is a
process). We are working on this.
BS

Alan Ruttenberg

unread,
Feb 11, 2010, 4:16:47 PM2/11/10
to bfo-discuss, piotrnowara
On Wed, Feb 10, 2010 at 4:39 PM, Piotr Nowara <piotr...@gmail.com> wrote:
> Dear all,
>
> I'd like to share some thoughts on the definitions of continuant and
> independent continuant:
>
> 1a) Continuant is defined as “entity [bfo:Entity] that exists in full
> at any time in which it exists at all, persists through time while
> maintaining its identity and has no temporal parts”.
>
> What does it mean „to exist in full”? How „not-existing in full” would
> differ from „existing in full”? Should I say that „I do exist in full
> during this very moment of my life, but the process of my life exists
> in 30%”?  What happens if the process of my life exists in 100%? What
> is the meaning of these kind of statements? How  this quantification
> of existing could be of any use?
>
> In my opinion this is an example of confusing existence with
> completeness. In fact the term „completeness” should be applied only
> to some occurents, more precisely to processes (maybe it could be also
> useful to describe events as something that is completed int the
> moment of occuring).

There's been a fair amount of traffic on the thread so I'll add only
one item. For an entirely different reason Barry and I have been
discussing "exists in full at any time in which it exists at all" part
of the definition of continuant and he agrees, on my urging, to remove
it.

A problem with it is that it makes it a requirement that anything that
doesn't exist in full at any moment of it's existence not be a
continuant. However there are a number of entities that he and others
consider to be continuant but which necessarily extend over (at least
a little) time. Some examples that have been discussed - speed,
acceleration, which necessitate at least an interval of time to be
defined, or the quality of being alive, which depend on there being
certain types of process instances that you are participating in.
Since the quality depends on those processes existing, and processes
don't exist in full at an instant of time, it would seem that the
clause would rule out "alive" as being a quality.

The other two clauses have been discussed here, seem sufficient to
convey the meaning of continuant, and, with discussion, hopefully make
some sense to you.

-Alan

Phillip Lord

unread,
Feb 12, 2010, 8:34:41 AM2/12/10
to bfo-d...@googlegroups.com

Alan Ruttenberg <alanrut...@gmail.com> writes:
> Some examples that have been discussed - speed, acceleration, which
> necessitate at least an interval of time to be defined

Speed, velocity and acceleration are defined over zero length intervals
of time.

> or the quality of being alive, which depend on there being certain
> types of process instances that you are participating in. Since the
> quality depends on those processes existing, and processes don't exist
> in full at an instant of time, it would seem that the clause would
> rule out "alive" as being a quality.

In this case, "alive" is being used to describe that the entity is
engaged in a process. This is compared, for example, colour or length
which doesn't necessitate this involvement; in BFOs model, objects have
colour even in the dark.

The slight worry here is a hierarchy duplication, as happens in
realizable entity; does every process need a quality which is the
quality that an object has when it's part of that process.


Phil

Barry Smith

unread,
Feb 12, 2010, 8:50:25 AM2/12/10
to bfo-d...@googlegroups.com
At 08:34 AM 2/12/2010, you wrote:


>Alan Ruttenberg <alanrut...@gmail.com> writes:
> > Some examples that have been discussed - speed, acceleration, which
> > necessitate at least an interval of time to be defined
>
>Speed, velocity and acceleration are defined over zero length intervals
>of time.

you cannot define speed at a time without reference also to
sufficiently short intervals


> > or the quality of being alive, which depend on there being certain
> > types of process instances that you are participating in. Since the
> > quality depends on those processes existing, and processes don't exist
> > in full at an instant of time, it would seem that the clause would
> > rule out "alive" as being a quality.
>
>In this case, "alive" is being used to describe that the entity is
>engaged in a process. This is compared, for example, colour or length
>which doesn't necessitate this involvement; in BFOs model, objects have
>colour even in the dark.
>
>The slight worry here is a hierarchy duplication, as happens in
>realizable entity; does every process need a quality which is the
>quality that an object has when it's part of that process.

The response to all such issues is: if the scientists find the need
for duplication here then we have to follow their lead; but then the
scientists will control the floodgates. Thus we certainly should not
have an axiom to the effect that every process need a quality which

is the quality that an object has when it's part of that process.

BS

Phillip Lord

unread,
Feb 12, 2010, 9:33:22 AM2/12/10
to bfo-d...@googlegroups.com
Barry Smith <phis...@buffalo.edu> writes:
> At 08:34 AM 2/12/2010, you wrote:
>>Speed, velocity and acceleration are defined over zero length intervals
>>of time.
>
> you cannot define speed at a time without reference also to sufficiently short
> intervals

Yes, the interval being zero in length.

>>In this case, "alive" is being used to describe that the entity is
>>engaged in a process. This is compared, for example, colour or length
>>which doesn't necessitate this involvement; in BFOs model, objects have
>>colour even in the dark.
>>
>>The slight worry here is a hierarchy duplication, as happens in
>>realizable entity; does every process need a quality which is the
>>quality that an object has when it's part of that process.
>
> The response to all such issues is: if the scientists find the need for
> duplication here then we have to follow their lead;

Whether this is a good answer depends on whether the need arises from
their requirements or the requirements of BFO.


> but then the scientists will control the floodgates. Thus we certainly
> should not have an axiom to the effect that every process need a
> quality which is the quality that an object has when it's part of that
> process.

Maybe. Ideally, I would look for ways to avoid this sort of maintenance
issue.

Phil

Alan Ruttenberg

unread,
Feb 12, 2010, 9:43:09 AM2/12/10
to bfo-d...@googlegroups.com
On 2/12/10, Phillip Lord <philli...@newcastle.ac.uk> wrote:
> Barry Smith <phis...@buffalo.edu> writes:
>> At 08:34 AM 2/12/2010, you wrote:
>>>Speed, velocity and acceleration are defined over zero length intervals
>>>of time.
>>
>> you cannot define speed at a time without reference also to sufficiently
>> short
>> intervals
>
> Yes, the interval being zero in length.

C'mon Philip. It's defined in the limit as the interval -> 0. You
can't have a limit without having an interval which is > 0 over which
to do the limit. That's the sense in which it depends on an interval
of time as opposed to being able to be defined at an instant.

-Alan

Alan Ruttenberg

unread,
Feb 12, 2010, 9:45:38 AM2/12/10
to bfo-discuss
On Fri, Feb 12, 2010 at 8:34 AM, Phillip Lord <philli...@newcastle.ac.uk> wrote:
> or the quality of being alive, which depend on there being certain
> types of process instances that you are participating in. Since the
> quality depends on those processes existing, and processes don't exist
> in full at an instant of time, it would seem that the clause would
> rule out "alive" as being a quality.

In this case, "alive" is being used to describe that the entity is
engaged in a process. This is compared, for example, colour or length
which doesn't necessitate this involvement; in BFOs model, objects have
colour even in the dark.

BFO doesn't (and shouldn't even in examples) say anything about color, but PATO does. 

I don't like the treatment of color in PATO. There are several things that are mixed together. There is the physical basis of transmitted or reflected color, which, proximately, has to do with the type and arrangement of atoms at in layer that light interacts with. This is a quality that does not have to do with time. When "color" refers to that, things have "color" even in the dark. 

Then there is the spectrum of light reflected by that surface when illuminated by a particular source illumination. This does, necessarily, have to do with time, but also has a dependency on the illumination, which is not recorded in PATO. In that sense of "color", things do not have "color" even in the dark.

Finally there is light emitted as atoms move between different excitation states, light emitted when matter is annihilated, and black body radiation, which isn't dependent on an illumination source, which is also dependent on time, and which has as it's proximate physical basis quantum mechanical properties of atoms. In that case the thing itself is the source of illumination and so dark or light is irrelevant.

The slight worry here is a hierarchy duplication, as happens in
realizable entity; does every process need a quality which is the
quality that an object has when it's part of that process.

Yes, I worry about that too. Barry's answer seems like the best answer for the moment, along with axioms in the ontology that connect them.

Something like:

has_quality alive equivalentClass participates_in (life_of self)

Might be able to do better at the instance level with some property-chain man-man[1] trick. 

-Alan

Wacek Kusnierczyk

unread,
Feb 12, 2010, 10:35:05 AM2/12/10
to bfo-d...@googlegroups.com

yes. the existence of a limit of a function (e.g., velocity or
acceleration as a function of time) at a point (e.g., zero) does not
require or imply the existence of a value of the function at that point.

vQ

Albert Goldfain

unread,
Feb 12, 2010, 11:49:02 AM2/12/10
to bfo-d...@googlegroups.com
Regarding the velocity discussion, I think there is a conflation of
"velocity" and "how velocity is calculated / expressed as a limit".
First a very quick aside:

There is a proof (probably only well known by certain mathematicians)
that 1 = 0.999... (read "point 9 repeating"). It runs as follows:
See http://en.wikipedia.org/wiki/0.999...

1/3 = 0.333... (read "point 3 repeating")

multiply both sides by 3:

3/3 = 0.999...

which is:

1 = 0.999...

So 1 and 0.999... (point 9 repeating) are just notational variants.
What is interesting about this result is that some people see the
right hand side 0.999... as "approaching 1" or "getting arbitrarily
close to 1, but never reaching it". But this is incorrect. Numbers
don't "approach" anything...they sit happily on the number line where
they live. And 0.999..., by the above proof, sits exactly at 1.

So why are people tempted to say 0.999... approaches 1? This is
probably because the series

S = {0.9, 0.99, 0.999, 0.9999, 0.99999, ...}

does approach 1. Or to put it more preciesly, each successive value
gets closer to 1. And it is the series S that people see when
carrying out, for example, the multiplication algorithm. But the
process to obtain this number is not the number itself.

Similarly, velocity is framed in terms of limits as

the limit as delta-t approaches zero of ...

again using "approaches". If *the* velocity is *the* limit, then
*the* velocity doesn't approach anything. It is stationary at some
point on the number line. Points on the number line are instants on
the timeline, so if your ontology honors instants as first-class
entities (and I understand that BFO does have instants), then
velocities can exist at instants.

The implication is here is that the oft quoted definition of velocity
as "instantaneous rate of change in position" is ontologically
incoherent because instants and change do not mingle. This leaves
open the possibility of seeing velocity as an infinte series (the only
option in an ontology with only temporal intervals) or seeing it as an
instantaneous quality (so "change" is wrong).

-Albert

Chris Mungall

unread,
Feb 12, 2010, 12:24:01 PM2/12/10
to bfo-d...@googlegroups.com, piotrnowara

perhaps the problem is not in the definition of continuant, but in the
attempts to shoehorn qualities into a C/O dichotomy?

My understanding is that the C/O distinction is one thing shared in
common by many upper ontologies. The distinction - essentially whether
part_of has an additional argument for time - also seems common across
these upper ontologies [disclaimer: IANAUO], even if it is worded in a
strange way in the current bfo definition. It's also a distinction
many non-ontologist biologists seem happy with (once they get over the
terminology). The proposed change to the bfo definition would weaken
this distinction rather than strengthen it.

What about:

* taking part_of/2 and part_of/3 as primitives
* x instance_of C at t iff exists y, part_of(y,x,t)
* x instance_of P iff exists y, part_of(y,x)
* merging IC and C
* making realizable, quality etc siblings of C and P in the hierarchy,
and forget about calling them continuants
* other abstractions that don't have physical or temporal parts would
also stand outside C and P - e.g. points, process boundaries


> The other two clauses have been discussed here, seem sufficient to
> convey the meaning of continuant, and, with discussion, hopefully make
> some sense to you.
>
> -Alan
>

Wacek Kusnierczyk

unread,
Feb 13, 2010, 6:20:53 PM2/13/10
to bfo-d...@googlegroups.com
Albert Goldfain wrote:
> Regarding the velocity discussion, I think there is a conflation of
> "velocity" and "how velocity is calculated / expressed as a limit".
>

i think there is a different conflation here. read on.

[...]

> Similarly, velocity is framed in terms of limits as
>
> the limit as delta-t approaches zero of ...
>
> again using "approaches". If *the* velocity is *the* limit, then
>

only *if* it is.

> *the* velocity doesn't approach anything.

which *the* velocity?


> It is stationary at some
> point on the number line. Points on the number line are instants on
> the timeline, so if your ontology honors instants as first-class
> entities (and I understand that BFO does have instants), then
> velocities can exist at instants.
>
> The implication is here is that the oft quoted definition of velocity
> as "instantaneous rate of change in position"

sheer curiosity, google:
"instantaneous rate of change in position": 7 hits
"rate of change in position": 951 000 hits


> is ontologically
> incoherent because instants and change do not mingle. This leaves
> open the possibility of seeing velocity as an infinte series (the only
> option in an ontology with only temporal intervals) or seeing it as an
> instantaneous quality (so "change" is wrong).
>
>

not 'change', but 'rate of change'.

anyway, i agree that approaching a limit should not be confused with the
limit itself. note the wording below (correct me if my use of initials
is incorrect):

[pl] Speed, velocity and acceleration are defined over zero length
intervals of time.

average velocity is defined as the change in location over an interval
of time.
instantaneous velocity is the value approached by average velocity in
the limit of the time interval approaching to zero.

which one do you call 'velocity', simply? if the first, then there is
no requirement for the intervals being zero length (and it would not
make sense if they were, as this would imply division by zero). if the
other, it's defined for time points (possibly considered zero-length
intervals).

[bs] you cannot define speed at a time without reference also to
sufficiently short intervals

yes, as much as the derivative of a function is defined in terms of the
difference between the values of the function at two points in the limit
of the difference between these points approaching zero.

[pl] Yes, the interval being zero in length.

which interval?

[ar] It's defined in the limit as the interval -> 0.

the *instantaneous* velocity is defined as the limiting value,
approached by the average velocity as the interval approaches zero.

[ar] You can't have a limit without having an interval which is > 0

over which to do the limit.

the limit is a value at a point, not over an interval. as albert says,
don't confuse the limit with how you arrive at it.

[ar] That's the sense in which it depends on an interval of time as

opposed to being able to be defined at an instant.

but now you must be talking about a series of average velocities over
shorter and shorter intervals starting at the same time point, rather
than about the instantaneous velocity *at* the time point.

vQ

Phillip Lord

unread,
Feb 15, 2010, 7:31:25 AM2/15/10
to bfo-d...@googlegroups.com


While I think we could have an very interesting discussion about the
maths, I don't think it's terribly relevant. Like it or not, science
worked out many years ago how to give meaning to things such as velocity
at a single point in time. I doubt that any discussion we have is going
to add to this, we should just accept it.

Even if you choose to have a long discussion and come to a "better"
conclusion, you will still have to live with the fact that the treatment
that has given meaning to velocity at a point in time has become the
standard definition. So, you will still need to describe the standard
definition and the relationship to your definition.

Finally, you seem to be suggesting that velocity and similar time based
concepts are a special case; they are not. The issue will come up in
many other areas. What is a slope of a line at a single point? What is
heat capacity of water at 20C? Even, what is the density of water at a
single point?

Velocity is $dr/dt$. A measurement over a non-infinitesimal time span is
an average.

Phil

Phillip Lord

unread,
Feb 15, 2010, 7:41:30 AM2/15/10
to bfo-d...@googlegroups.com, piotrnowara
Chris Mungall <c...@berkeleybop.org> writes:
> perhaps the problem is not in the definition of continuant, but in the
> attempts to shoehorn qualities into a C/O dichotomy?
>
> My understanding is that the C/O distinction is one thing shared in common by
> many upper ontologies. The distinction - essentially whether part_of has an
> additional argument for time - also seems common across these upper
> ontologies [disclaimer: IANAUO], even if it is worded in a strange way in the
> current bfo definition. It's also a distinction many non-ontologist
> biologists seem happy with (once they get over the terminology). The proposed
> change to the bfo definition would weaken this distinction rather than
> strengthen it.
>
> What about:
>
> * taking part_of/2 and part_of/3 as primitives
> * x instance_of C at t iff exists y, part_of(y,x,t)
> * x instance_of P iff exists y, part_of(y,x)
> * merging IC and C
> * making realizable, quality etc siblings of C and P in the hierarchy, and
> forget about calling them continuants
> * other abstractions that don't have physical or temporal parts would also
> stand outside C and P - e.g. points, process boundaries


Chris

If I understand your notation, most of this seems sensible. However,
part of the problem with qualities is not that they, themselves, are
continuants. It's that qualities can only be borne by ICs and, more,
that they have to be borne by at least one IC.

It's this that leads to the conclusion that a spatial region can't have
a length. Strictly, it leads to the conclusion that

spatial region bears length

is not consistent; of course, this doesn't mean that we can't create
another relationship, say "blah" and say

spatial region blah length

but now we have must have some IC that this length must inhere in, and
it's not at all clear what this IC is.

We get the same issue, but in a slightly different way, with GDCs.
Intuitively, you might want to say

Data bearer size_in_kB

where size_in_kB is a quality. But you can't, because Data is a GDC.

Essentially, with BFO as it stands, the only solution that I can see is
duplicate the design pattern that we have in quality wrt to independent
continuants. We would need a relationship and a hierarchy for the
properties of both processes and dependent continuants. To me, this is
all fairly complex.

Phil

Michel Dumontier

unread,
Feb 15, 2010, 8:40:22 AM2/15/10
to bfo-d...@googlegroups.com, piotrnowara
There is little doubt in my mind that even information content entities have attributes. We'll need relations to link IC or ICE's *directly* with measured/observed attributes.

m.
 
--
You received this message because you are subscribed to the Google Groups "BFO Discuss" group.
To post to this group, send email to bfo-d...@googlegroups.com.
To unsubscribe from this group, send email to bfo-discuss...@googlegroups.com.
For more options, visit this group at http://groups.google.com/group/bfo-discuss?hl=en.




--
Michel Dumontier
Associate Professor of Bioinformatics
Carleton University
http://dumontierlab.com

Phillip Lord

unread,
Feb 15, 2010, 9:10:04 AM2/15/10
to bfo-d...@googlegroups.com
Michel Dumontier <michel.d...@gmail.com> writes:
>> Essentially, with BFO as it stands, the only solution that I can see is
>> duplicate the design pattern that we have in quality wrt to independent
>> continuants. We would need a relationship and a hierarchy for the
>> properties of both processes and dependent continuants. To me, this is
>> all fairly complex.
>>
>> Phil
>>
>>
> There is little doubt in my mind that even information content entities have
> attributes. We'll need relations to link IC or ICE's *directly* with
> measured/observed attributes.


Yep. The issue is whether the attributes for dependent continuant,
independent continuants and processes are disjoint, or whether they
overlap.

Phil

Reply all
Reply to author
Forward
0 new messages