search

obo-behavior / behavior-ontology

Neuro Behaviour Ontology: an ontology for human and animal behaviour processes and behaviour phenotypes
25 stars 15 forks source link

Sub-branch review: Stress related behavior #123

Closed DitchingIt closed 1 year ago

DitchingIt commented 1 year ago

Method

With reference to the principles and guidelines named in issue #122, I have reviewed the 17 subclasses in the stress related behavior sub-branch of the behavior process branch of NBO. I do not propose to go into such detailed explanations in any future sub-branches, but this is an opportunity to demonstrate my method.

Apologies

I have made changes to a fork in my repository using Protege for a pull request in due course, and look forward to modifying it on the basis of discussions here. This is my first time doing all this, so apologies in advance for doing it wrong or appearing to criticise anyone with the changes I have suggested.

Analysis

I have split my notes up into a task list below, in case they need to be broken down into individual issues.

NBO Style and Conventions

For general discussions about NBO style and conventions which refer to the S1 bullet below (i.e. not specific to a stress related class definition), I am copying it into issue #122 and suggest continuing it there.

pmidford commented 1 year ago

I've started reviewing these, and the first three seem to be based on physiology rather than behavior. In particular, stress related behavior seems to be defined in terms of mechanism (allostatic overload). The next two are defined in terms of the general adaptation syndrome model. If defining this in terms of mechanism is the best way to do so, then I'm fine, but maybe we need a physiology ontology (or GO additions?) to handle these.

The term id for agonistic alarm indicates it came from the ABO, and it does have a definition there: "Behavior signaling awareness of and or fear of danger, triggered by agonistic actions or postures of a conspecific rather than a predator." I don't have a source for that, maybe @aclark-binghamton-edu, who worked with a collaborator to develop the ABO definitions from established reference sources could find it.

I'll keep looking through these. I do appreciate your effort in proposing these definitions, I think the role of physiology and its relation to behavior, in particular in NBO, is something worth working through.

DitchingIt commented 1 year ago

I appreciate your reviewing these. But fundamentally, I'm asking if my labelling and defining method is valid; these are more like pedantic examples: we may never want to go through with these exact redefinitions. My main effort is currently going into producing a theoretically sound behaviour model with the prospect of totally rebooting the NBO - look out for that soon hopefully. Then if that won't float, I'd come back in earnest to the sub-branch reviews like the stress one above.

The physiology question is exactly the kind of observation I'm most looking for though:

  1. First of all, I am searching for ways to stop using the word 'behaviour' every time we label or define a behaviour.
  2. Second I'm starting with what currently exists in NBO, and I guess the root of stress is physiological.
  3. I think that we need willingly to skim the surface of physiology and anatomy in places, not least where functional anatomy/ morphophysiology abound.
  4. I've made a more recent attempt at defining the acute stress response, and pulled back a bit from the example above, to: "The second phase of the first stage of adaptation that activates the fight-or-flight response." Keeping you in mind will reduce slippage in that direction: it's valuable to be kept on one's toes!
  5. Where we have domains (like stress maybe) which seem to be full of physiology, or really to be all about triggers, with little definite or specific in the way of behaviour, maybe we need to do some pruning...
  6. Regarding agonistic alarm, it would need to look/sound different to predator alarm, but this is a good direction to be searching in for a source; I really don't want to end up with lists and lists of triggers for exactly the same behaviour though.
  7. The bottom line though, is that I may be getting it wrong and more perspectives like this will be invaluable.
pmidford commented 1 year ago

The labeling and defining method looks good to me, it's certainly an improvement over the haphazard state of things. This and the other issues have convinced me to review the pages and blog posts you linked, both for NBO and what I have been doing with the spiders. I will wait to see what you come up with in a behavior model. You might want to look at ABO as well - it's an ontology a group of animal behaviorists developed about 15 years ago, there's a copy in this project.

DitchingIt commented 1 year ago

You might want to look at ABO as well - it's an ontology a group of animal behaviorists developed about 15 years ago, there's a copy in this project.

Will do - thanks.

aclark-binghamton-edu commented 1 year ago

To enter this conversation as one of the animal behaviorists/behavioral ecologists who developed the ABO and then worked on the merger--and forgive me if some of what I say below is obvious: Definitions in NBO were often missing when we were merging the two ontologies and I did write many, where clear and appropriate, to follow ABO definitions; every ABO term had a definition. However, before ABO was merged into NBO, I reformatted ABO definitions or modified ABO terms (e.g. noun form vs verb or gerund form) to follow the patterns set up by NBO. That is where the "Behavior of..." or "xxx Behavior" comes from. In NBO, a behavior that ABO termed "sucking" was termed in NBO "sucking behavior". Furthermore, where ABO had, e.g. "urinate", NBO 's possible equivalent was "Behavioral Control of urination".

I still have my spreadsheets showing equivalencies or assumed equivalencies of terms, as well as definitions for ABO terms (and thus for equivalent NBO terms). Reasoning across the two hierarchies was sometimes difficult because parent-child relationships were often different. There were many terms in NBO that did not apply to ABO, which emphasized normal or ecologically relevant behaviors. It attempted to describe what was observable visually or auditorily (e.g. for vocalizations). It also left out aspects that could be added as a modifier for a particular study or context: quality or strength of behavior, by male vs female, etc. And of course, it had to stop at a fairly high level of description to accommodate many different species.

An ontology emphasizing the ABO goals would be most welcome. I am using the ABO as represented in its pre-merger form to organize my American Crow Ethogram; this is being used and will be made further available to those working with other crow species.

On Mon, Jan 16, 2023 at 3:31 PM Ditch Townsend @.***> wrote:

You might want to look at ABO as well - it's an ontology a group of animal behaviorists developed about 15 years ago, there's a copy in this project.

Will do - thanks.

— Reply to this email directly, view it on GitHub https://github.com/obo-behavior/behavior-ontology/issues/123#issuecomment-1384525018, or unsubscribe https://github.com/notifications/unsubscribe-auth/ADUXTFQSPHMHQ3SN53XBJATWSWVZVANCNFSM6AAAAAATUAV5BE . You are receiving this because you were mentioned.Message ID: @.***>

-- Anne B. Clark, Ph.D. Biological Sciences Binghamton University Binghamton, NY 13902-6000

(607) 222-0905 (cell) (607) 777-6521 (fax-Dept office) (607) 777-2438 (Biol Sci office)

DitchingIt commented 1 year ago

Thanks Anne. I need to get into the ABO stuff in detail, but I am certainly leaning in your direction by the sound of it.

DitchingIt commented 1 year ago

I will wait to see what you come up with in a behavior model.

@pmidford I've posted my model at #126 and look forward to feedback. It's not make-or-break for me, but I wanted to get it out there as a first option.

dosumis commented 1 year ago

stress related behavior ORIGINAL (definition): "G1aBehavior related toG6.1a how the bodyG5.1a reacts to a stressorG7a (a stimulusG3.1.2a that causes stressG7b),G6.2a real or imagined."G5.1b,P6a,S1a,S11a [wikipedia: Stress_(biological)]G2a,G2b

CITATION: wikipedia: Stress_(biological) CROSS REFERENCE (xref): None SUBCLASS OF: behavior process

PROPOSED (definition): A processG2c,S3.1.1a thatP6a responds toG3.1.3a allostatic overload.G5a,G6.1a,G7c CITATION: wikipedia: Allostasis#Types CROSS REFERENCE (xref): GO:0006950 response to stressXa SUBCLASS OF: behavior process

I've started reviewing these, and the first three seem to be based on physiology rather than behavior. In particular, stress related behavior seems to be defined in terms of mechanism (allostatic overload). The next two are defined in terms of the general adaptation syndrome model. If defining this in terms of mechanism is the best way to do so, then I'm fine, but maybe we need a physiology ontology (or GO additions?) to handle these.

We absolutely need to keep behaviour in here. Stress is a physiological state that triggers many non-behavioural processes, the behavioural term is for behaviours in response to stress. The proposed definition makes it indistinguishable from the GO term you xref when it should be a subclass.

IIRC, ABO maintained a clean division between observed behaviour and functional/causal interpretation. We should still support this - after all, interpretation of function/cause may change with no change to the observation. In merging with NBO, we allowed for behavioural classes that included cause/function. These should be considered OWL compound classes with behaviour as the genus and slots for cause/function. When we merged in NBO, we discussed formalising such patterns, but in the absence of funding/resources, this never happened.

DitchingIt commented 1 year ago

Really helpful, although I don't understand xrefs technically. My knowledge of OWL is also near zero: doubtless, "OWL compound classes with... slots..." is what I would suggest if I knew it was what I meant (and don't worry, I really am not being sarcastic). For the forseeable future, I see my time and motivation as a key resource for NBO and I hope you can help me, alongside Nico, to do what I can technically. But I certainly agree that the ABO is a good direction to look in.

dosumis commented 1 year ago

I'd forgotten that we have a partial write-up of the workshop where we decided to merge ABO and NBO. This outlines the approach pretty well I think. https://docs.google.com/document/d/1eSCjQfR5BujtUYw4W2pYOL6rhPPMycJd8u95X1-5U7Q/edit#

Apologies for not sharing earlier.

Unfortunately I think this clashes with many of your suggestions. e.g.

First of all, I am searching for ways to stop using the word 'behaviour' every time we label or define a behaviour.

I don't think we can avoid appending 'behaviour' in the case of general grouping classes - function/cause + behaviour.

I also think we need to step back a bit an detail use cases for NBO before we attempt anything radical. This is particularly important when it comes to assertions of redundancy. Apparently redundant terms may not really be redundant when we look at use cases.

Use case 1: Ethology - ABO approach

Observations need to be recorded separately from interpretations (cause/function). I think that the best way to do this is with separate terms for function/cause (as ABO).

This may also be appropriate for observational work on model organisms (e.g. home cage experiments in mice or FlyBowl in Drosophila)

Use case 2: Behavioural studies of model organisms centred on behavioural testing paradigms.

Here we mostly want to use phenotype terms, but need a behaviour ontology to defined phenotypes. A mix of observational behavioural classes and those combined with some function is needed for non-controversial, well studied paradigms. e.g. Drosophila biologists would expect to see the dipteran jump response classified as an escape response. The compound term approach in the above doc supports this.

There are some tricky boundary problems around mental events & processes vs behaviour (e.g. learning and memory). NBO glosses these, which may be fine given that we need these classes and upper ontology fussiness is likely to get in the way of supporting this use case.

Use case 3: Clinical use cases

Lots more expert input is needed here. We need to support classes for behaviours caused by internal mental state or external stimulus (needed for other animals too - just much more so here). e.g. - maybe we need to broaden, not narrow the definition here so that it more explicitly covers cases where the stimulus is internal (maybe it does already cover this in an triggering stimulus may be a mental state):

fear-related behavior ORIGINAL (definition): "G1aAn emotional behavior related toG6.1a a feeling of uneasiness orG3.1.3b nervousness triggered by a specified triggeringG7d stimulus such as pain or the threat of danger."G2h,G3.1a,G5c,G9.2c,P6a,S11a [NBO:GVG]

CITATION: [NBO:GVG]G2d SUBCLASS OF: fear/anxiety related behavior

PROPOSED (definition): An emotional responseG2c,G3.1b,G3.1.1a thatP6a is triggered by a real present danger.S5a

Suggestion:

We collect use-cases, with examples, in the README of this repo and use these to assess proposed changes.

DitchingIt commented 1 year ago

Thanks. I think I understand the term 'use case' but can you give me some examples and where to find them, so I don't get the wrong end of the stick? I suspect 'we' means 'me'.

DitchingIt commented 1 year ago

Thanks. I think I understand the term 'use case' but can you give me some examples and where to find them, so I don't get the wrong end of the stick? I suspect 'we' means 'me'.

Sorry - it was right in front of me and I didn't notice (Use case 1, Use case 2, Use case 3)!

DitchingIt commented 1 year ago
  1. The original stress-related behavior definition review I did at the top of this issue was less about stress than about conventions. It has informed, and been taken over by #122
  2. The use case discussion has been taken over by #127