The Behaviour Change Technique Ontology: Transforming the Behaviour Change Technique Taxonomy v1

Background The Behaviour Change Technique Taxonomy v1 (BCTTv1) specifies the potentially active content of behaviour change interventions. Evaluation of BCTTv1 showed the need to extend it into a formal ontology, improve its labels and definitions, add BCTs and subdivide existing BCTs. We aimed to develop a Behaviour Change Technique Ontology (BCTO) that would meet these needs. Methods The BCTO was developed by: (1) collating and synthesising feedback from multiple sources; (2) extracting information from published studies and classification systems; (3) multiple iterations of reviewing and refining entities, and their labels, definitions and relationships; (4) refining the ontology via expert stakeholder review of its comprehensiveness and clarity; (5) testing whether researchers could reliably apply the ontology to identify BCTs in intervention reports; and (6) making it available online and creating a computer-readable version. Results Initially there were 282 proposed changes to BCTTv1. Following first-round review, 19 BCTs were split into two or more BCTs, 27 new BCTs were added and 26 BCTs were moved into a different group, giving 161 BCTs hierarchically organised into 12 logically defined higher-level groups in up to five hierarchical levels. Following expert stakeholder review, the refined ontology had 247 BCTs hierarchically organised into 20 higher-level groups. Independent annotations of intervention evaluation reports by researchers familiar and unfamiliar with the ontology resulted in good levels of inter-rater reliability (0.82 and 0.79, respectively). Following revision informed by this exercise, 34 BCTs were added, resulting in the first published version of the BCTO containing 281 BCTs organised into 20 higher-level groups over five hierarchical levels. Discussion The BCTO provides a standard terminology and comprehensive classification system for the content of behaviour change interventions that can be reliably used to describe interventions. The development and maintenance of an ontology is an iterative and ongoing process; no ontology is ever ‘finished’. The BCTO will continue to evolve and grow (e.g. new BCTs or improved definitions) as a result of user feedback and new available evidence.

reliably apply the ontology to identify BCTs in intervention reports; and (6) making it available online and creating a computer-readable version.

Results
Initially there were 282 proposed changes to BCTTv1.Following firstround review, 19 BCTs were split into two or more BCTs, 27 new BCTs were added and 26 BCTs were moved into a different group, giving 161 BCTs hierarchically organised into 12 logically defined higher-level groups in up to five hierarchical levels.Following expert stakeholder review, the refined ontology had 247 BCTs hierarchically organised into 20 higher-level groups.Independent annotations of intervention evaluation reports by researchers familiar and unfamiliar with the ontology resulted in good levels of inter-rater reliability (0.82 and 0.79, respectively).Following revision informed by this exercise, 34 BCTs were added, resulting in the first published version of the BCTO containing 281 BCTs organised into 20 higher-level groups over five hierarchical levels.

Discussion
The BCTO provides a standard terminology and comprehensive classification system for the content of behaviour change interventions that can be reliably used to describe interventions.The development and maintenance of an ontology is an iterative and ongoing process; no ontology is ever 'finished'.The BCTO will continue to evolve and grow (e.g.new BCTs or improved definitions) as a result of user feedback and new available evidence.

Introduction
Descriptions of behaviour change interventions vary widely, undermining the ability to synthesise evidence or replicate interventions for evaluation or implementation.This is a barrier to accumulating evidence about intervention effectiveness and thus making recommendations for research, policy, and practice.It also hinders developing more effective interventions.For this reason, a method for specifying intervention content was developed in the form of a structured taxonomy of behaviour change techniques, the Behaviour Change Technique Taxonomy v1 (BCTTv1) (Michie et al., 2013;Michie et al., 2015).This paper describes the development and evaluation of the next generation of intervention description technology, the Behaviour Change Technique Ontology (BCTO).
Behaviour change techniques (BCTs) are defined as the smallest parts of the content of a behaviour change intervention that are observable, replicable and on their own have the potential to bring about behaviour change (Michie et al., 2021a).The BCTTv1, developed with the input of 400 experts from around the world, comprises 93 BCTs, organised in 16 higher-order groupings based on cluster analysis of connections made by experts (Michie et al., 2013;Michie et al., 2015).It provides a standardised, shared language to describe the 'active ingredients' of an intervention.Resources were developed to support the use of BCTTv1, including a smartphone app (http://bit.ly/BCTsappGoogle;http://bit.ly/BCTsappApple), online training to guide the identification of BCTs in published papers (http://www.bct-taxonomy.com/), a database of studies of interventions coded using BCTTv1 (www.bct-taxonomy.com/interventions) and the Theory and Techniques Tool to link BCTs to their hypothesised mechanisms of action (https://theoryandtechniquetool.humanbehaviourchange.org) (Michie et al., 2021a).
BCTTv1 has been widely applied internationally, reported in more than 5000 published studies.These cover intervention design and evaluation, evidence synthesis and implementation of behaviour change interventions in research and practical settings.Using meta-regression, it has been applied to investigate the effectiveness of individual or group-based behaviour change interventions across a wide range of populations, settings, and behaviours (e.g., Carraça et al., 2021;Michie et al., 2018).In combination with frameworks such as the Behaviour Change Wheel (Michie et al., 2011;Michie et al., 2014), BCTTv1 has enabled a structured and systematic method for designing and evaluating interventions.
The BCTTv1 was intentionally named 'v1' to signal that developments to the taxonomy would be needed as the field advanced and feedback from users accumulated.To inform the improvement of the BCTTv1, we brought together user feedback from six sources (Corker et al., 2023).These were the BCT website, a user survey, researchers and experts involved in the Human Behaviour-Change Project, an interview-based consultation exercise of researchers and other users, relevant published research reports and other classification systems of BCTs.This feedback suggested a need to extend the BCTTv1, improve the labels and make the definitions more precise, and develop the structure to be more flexible, extensive, and multi-level.
Structures for representing knowledge by defining classes of entities (anything that exists in the universe) and their relations are called ontologies (Arp et al., 2015).In this paper, the terms 'entity' and 'class' are often used interchangeably to refer to entities represented in an ontology.For the definitions of technical terms used in this paper (in bold and italicised), see the glossary in Table 1.Entities and their relations are defined to represent their essential properties in such a way that they are uniquely and fully specified and assigned a unique label.This enables data to be computer-readable, and thus allows computational analysis of large amounts of complex data.'Computer-readable' refers to a data structure that can be directly processed by computers, which is not possible for natural language text.Representation of information in computerreadable data structures allows computers to process information unambiguously and effortlessly.According to the Open Biological and Biomedical Ontology (OBO) Foundry principles of good practice for developing ontologies, ontologies must be made available in a common formal language in an accepted concrete syntax (Principle 2), which is typically the Web Ontology Language (OWL).This is necessary to investigate how behaviour change intervention components interact in producing effects and explanations of variation across, for example, populations, settings, and behaviours.
Ontologies offer a more comprehensive and expressive way of representing information than taxonomies (Hastings, 2017).For example, they can link BCTs to other intervention features such as their delivery, mechanisms of action and target behaviours, and context entities, such as population and setting.Ontologies also provide an effective method for connecting and accumulating knowledge across topic domains and academic disciplines (i.e.provide 'interoperability').Because they enable reporting in a clear, structured and transparent way, ontologies support clear communication and collaborative sharing of data between researchers and others.A further advantage of ontologies is that they are not static; they are designed to be added to and amended as new information accumulates from the use of the ontology and from scientific and intellectual advances.
The development of a Behaviour Change Technique Ontology (BCTO) was informed by 282 feedback comments on BCTTv1 that suggested the need for additional BCTs, amendments to

Annotation
Process of coding, or tagging, selected parts of documents or other resources to identify the presence of ontology entities.Michie et al. (2017) Annotation guidance manual Written guidance on how to identify and tag pieces of text from intervention evaluation reports with specific codes relating to classes in the ontology, using for example EPPI-Reviewer software.Michie et al. (2017) Basic Formal Ontology (BFO) An upper-level ontology specifying foundational distinctions between different types of entity, such as between continuants and occurrents, developed to support integration, especially of data obtained through scientific research.

Arp et al. (2015)
Behaviour Change Intervention (BCI) An intervention that has the aim of influencing human behaviour.OWL is designed to be interpreted by computer programs and is extensively used in the Semantic Web where rich knowledge about web documents and the relationships between them are represented using OWL syntax.
https://www.w3.org/TR/owl2-quick-reference/ labels and definitions of specific BCTs, amendments to the groupings, and general improvements to increase clarity.The work was conducted as part of developing the overarching Behaviour Change Intervention Ontology (Michie et al., 2021b), part of the Human Behaviour-Change Project (Michie et al., 2020).The Behaviour Change Intervention Ontology (BCIO) represents behaviour change interventions and their evaluations.The BCIO consists of an upper level with 42 entities, one of which is behaviour change intervention content (BCI content), which includes the entity "behaviour change technique" (see Figure 1).The BCIO covers the behaviour being targeted, how BCTs are delivered, e.g., their mode of delivery (Marques et al., 2021), schedule, style of delivery (Wright et al., 2023), and source (Norris et al., 2021), the context (setting and population) in which they are delivered (Norris et al., 2020), and the mechanisms through which they produce behavioural changes (Schenk et al., 2023).Note that the BCTO is distinguished from the BCIO (i.e., the BCTO forms part of the larger BCIO).
The BCIO is intended to be used by both humans and computers.Humans can use the ontology to describe and report interventions, synthesise evidence (e.g. through literature annotation), and develop interventions, to name a few possible use cases.In computer-readable form, the ontology is represented in OWL, which can be parsed and used in several different downstream applications.For example, we can harness the power of computation in terms of Machine Learning and Artificial Intelligence to automate the extraction of information from reports (West et al. 2023), and build systems that can predict behaviour in novel scenarios (Hastings et al. 2023).Ontologies provide an interface between computers and humans, that is, computers can better support human tasks that require understanding of a domain, coupled with integration of content at scale and speed.

Aim
This study aimed to develop an open-access, computer-readable Behaviour Change Technique Ontology (BCTO) that can be reliably used to describe the content of behaviour change interventions.

Ethical statement
Ethical approval was granted by the University College London's ethics committee (CEHP/2016/555).

Design
The development of the Behaviour Change Technique Ontology (BCTO) consisted of six iterative steps based on methods developed in the Human Behaviour-Change Project (Wright et al., 2020).Feedback from users of the BCTTv1 was analysed (Corker et al., 2023), BCT labels and definitions were rewritten to be consistent with ontological definitions (Michie et al., 2019), BCTs were organised in a logical ontological structure, expert feedback was incorporated and inter-rater reliability of BCTs in annotated intervention reports was assessed.
Step Step 2: Changes to BCTs: labels and definitions Step 1 recommendations were applied to each BCT label and definition, and changes made to aid clarity and specificity, where necessary, by authors EC and MJ.The revised BCTs were reviewed and amended where necessary by five of the study behavioural science experts (EC, MM, MJ, SM, and RW) and one ontology expert (JH).
Step 3: Structuring the BCTO as an ontology Step 1 recommendations were applied to each group label and BCTs within each group by EC and MM.Changes to add clarity were proposed by EC and MJ.The full set of revised group labels and BCTs within each group were reviewed and amended where necessary by the full team (EC, MM, MJ, SM, RW and JH).All BCTs were reviewed to ensure they had a hierarchical relationship ("is_a" in ontological terms) with their parent class (i.e., meaning the BCT is a subclass of the higher-level group it belongs to).New BCTs were discussed in relation to which group they belonged to by EC, MM, MJ and SM prior to the final team review to ensure that each group was inclusive (i.e., they contained BCTs with a common active content element) and exclusive (i.e., the BCTs within each group did not belong in any other group).
Step 4: Expert stakeholder review 21 expert stakeholders (17 behavioural scientists and 4 ontology experts) reviewed the BCTO resulting from Step 3 so that the ontology reflected broader scientific consensus about BCTs as well as meeting the requirements of ontology users (Wright et al., 2020).Behavioural scientists were recruited from a database of those expressing willingness to participate as expert reviewers for studies conducted at the UCL Centre for Behaviour Change.To be eligible, participants were required to have a doctoral level degree in behavioural science or a related discipline.We excluded those who were close collaborators of the BCTO's lead developers, i.e., had co-authored a publication in the previous three years or worked for the same institution.We purposively sampled to ensure geographical diversity and a range of career stages (from early career postdoctoral researchers to full professors or equivalent).Ontology experts were suggested by the Human Behaviour-Change Project's ontology expert (JH).Prospective participants were sent an invitation and study information sheet.Those willing to participate in the study were sent a link to an online questionnaire (https://osf.io/2gs9b)(West et al., 2020), along with the BCTO displayed in both spreadsheet and diagram forms.Experts were asked to review all groups of BCTs (and the individual BCTs within that group) one at a time, taking an estimated 2.5 hours and were paid an honorarium for doing this.The expert review was conducted using Qualtrics™.
Of the 17 behavioural scientists invited to participate, eight completed the review.In addition, two behavioural scientists developing a physical activity ontology provided feedback.Three of the four invited ontology experts completed the review.The 13 providing feedback worked in institutions based in the United Kingdom (n=4), Belgium (n=3), South Africa (n=1), Canada (n=1) and USA (n=4).
Participants were presented with the label of a single higher-level group from the BCTO and all the BCTs within that group.
For each BCT, participants were asked to indicate whether any labels or definitions needed refining and, if so, to suggest alternatives.They were asked for additional BCTs and for any other comments about the BCTO.Following a conference presentation, we received feedback from the Habit Special Interest Group of the European Health Psychology Society on the seven BCTs that were at that point in the "habit BCT" group.All feedback was discussed by the research team and led to revising BCT labels or definitions, rearranging BCT groupings, removing or adding BCTs or providing explanations for not revising.
Step 5: Inter-rater reliability of annotations using the BCTO Inter-rater reliability of annotations using the BCTO was assessed in two ways.First, two researchers involved in the development of the BCTO independently annotated 50 papers from Cochrane reviews (25 on smoking cessation and 25 on physical activity).This number was selected as it gives a 10-15% margin of error around the estimated percentage agreement between coders (Gwet, 2014;Wright et al., 2020).Annotations followed an annotation guidance manual (https://osf.io/mwv2c)(West et al., 2020).From this set of annotations, any necessary changes to the manual and labels or definitions of the BCTs were made.In the second assessment of inter-rater reliability, two behaviour change experts experienced in annotating behaviour change intervention reports but with no prior knowledge of the ontology, independently annotated a randomly selected sample of 50 randomised controlled trials from a database of papers coded using the Behaviour Change Techniques Taxonomy v1 (http://www.bct-taxonomy.com/interventions).
Inter-rater reliability was assessed using Krippendorff's Alpha (Hayes & Krippendorff, 2007) calculated using version 1.0.0 of the Automation Inter-Rater Reliability script developed for the Human Behaviour-Change Project (Finnerty & Moore, 2020).The research team made additional changes to the BCTO based on the issues arising from inter-rater reliability testing, as well as from a final revision of the consistency between class, definitions and labels.

Results
The results from each step of the BCTO development are presented in Figure 2 and described below.
Step 1: Extract and synthesise feedback on the BCTTv1 A total of 282 comments from the feedback exercises and published reports were received and used for review.These were organized into four categories: i. 32 comments containing 47 suggestions for new BCTs, ii.92 comments related to amendments to labels and definitions of specific BCTs, iii. 9 comments related to amendments to the groupings, and iv.17 comments containing suggestions for general improvements.Changes resulting from these recommendations are provided in Steps 2 and 3 (for full details also see Corker et al., 2023).
Step 2: Changes to BCTs: labels and definitions First, the definition of the term 'behaviour change technique' was updated to comply with ontological terms.The definition agreed by the research team was " A planned process that is the smallest part of BCI content that is observable, replicable and on its own has the potential to bring about behaviour change" (Michie et al., 2021a).
Second, each BCT was amended in the following ways: • BCT labels were revised so that each clearly aligned to a specific BCT definition.
• BCT labels and definitions were revised to ensure clarity of the active content, that is, exactly what process the BCT is describing.
• Technical and theory-specific language was removed from labels and definitions to allow for understanding across disciplines.
• 26 BCTs were moved to a group that better reflected the active content described.For example, 'demonstration of the behaviour BCT' was moved from the BCTTv1 cluster 'comparison of behaviour' to the BCTTv1 cluster 'shaping knowledge' (https://osf.io/j5wgb)(West et al., 2020).
• 27 new BCTs were agreed.These consisted of suggestions for 22 new BCTs from Step 1 and a further five agreed on during review meetings.For example, 'facilitate integration of behaviour goals BCT' was added to the 'goal directed BCT' group (https://osf.io/j5wgb)(West et al., 2020).
Step 3: Structuring the BCTO as an ontology • The numerical indicators of the groupings of BCTs were removed to better reflect ontological nomenclature -i.e., that the primary labels of classes in ontologies should not ordinarily contain numeric codes.Numeric codes are instead captured in the unique identifiers or where needed as associated annotations.For example, the unique identifier for 'Set behaviour goal BCT' is BCIO:007003.
• The BCTO was organised into a three-level classification hierarchy (https://osf.io/3tekn)(West et al., 2020).An individual BCT (e.g., 'Set behaviour goal BCT') was classified at the lowest level of the hierarchy (i.e., Level 3), with its parent class one level up in the hierarchy (e.g., 'Goal setting BCT' at Level 2).The highest-level of the ontology (i.e., Level 1) contains the parent classes of BCTs that share a common active content (e.g., 'Goal directed BCT') • All parent classes names were changed to better reflect the common active content described by each BCT within the group.
• A definition for each parent class was added to ensure clarity regarding the nature of the BCTs within the group.
• Four higher-level groups (comparison of outcomes, scheduled consequences, self-belief and covert learning) were removed as it was agreed that the active content described by the BCTs within these groups were set out clearly in definitions for other groups.• 13 lower-level parent classes were added across six groups to aid specificity.For example, the lower-level parent classes 'goal setting BCT' and 'goal implementing BCT' were added to the higher-level group 'goal directed BCT' (see https://osf.io/a6bwf)(West et al., 2020).
• The start of each BCT definition was amended to ensure that the definition included the label of its parent class, for example, the start of the definition for 'goal setting BCT' is 'a goal directed BCT that changes behaviour by….',where 'goal directed BCT' is the label for the higher-level group in which 'goal setting BCT' is placed.
Step 4: Expert stakeholder review The experts provided 326 comments on the ontology via the online survey.The EHPS Habit SIG made 11 recommendations regarding the eight BCTs in the "Habit BCT" group (see https://osf.io/2jwqz for responses to comments collected via the online survey and https://osf.io/6vhp4for responses to the EHPS Habit SIG's recommendations (West et al., 2020)).
In response to experts' feedback that it was unclear whether the definition of a BCT applied to both self-enacted behaviour change and behaviour change interventions delivered by a separate intervention source (e.g., health care professional), the definition of a BCT was changed to "A planned process that is the smallest part of behaviour change intervention content that is observable, replicable and on its own has the potential to bring about behaviour change in oneself or other people" (added text italicised.)We also removed mention of "the intervention source" from most definitions to clarify that BCTs could be delivered either by others or self-enacted.
Feedback from the ontology experts led to "that changes behaviour" being removed from the first part of the definition of all BCTs.For example, "goal setting BCT"'s definition was revised from "A goal-directed BCT that changes behaviour through goal setting" to "A goal-directed BCT that sets goals".This reflects a principle of ontological definitions that they should reflect what is always true about members of a class, and behaviour change techniques do not always lead to changes in behaviour.To make it clearer that BCTs were processes (i.e., things that take place over time), we updated BCT labels to include a verb where possible.
A number of stakeholder comments pointed out that many BCTs have more than one potential mechanism of action.Therefore, BCT groups where the only shared feature of the BCTs was their hypothesised mechanism of action (e.g., habit BCT group, personal resources BCT group) were regrouped according to the type of process involved in the BCT itself (e.g., "advise specific behaviour BCT" or "suggest different perspective on behaviour BCT" groups).In response to comments that some of the "creating consequences", "reward" and "incentive" BCTs had confusing definitions, we revised the organisation of these BCTs and added a number of new BCTs where their absence had been noted.
Expert feedback led to 25 BCTs being removed from the ontology (e.g.'prompt problem solving' was removed as it could not be distinguished from 'problem solving BCT') and 111 BCTs added (e.g.'set measurable behaviour goal BCT' was added as a child class of 'set behaviour goal BCT', as not all goals are measurable).All 137 BCTs retained from the Step 3 ontology had revised definitions, 87 had revised labels (e.g.'graded tasks BCT' was amended to 'set graded tasks BCT' to include the verb) and 73 had a revised parent class (e.g.'monitor emotional consequences BCT' was moved from the 'awareness of consequences BCT' group to the 'monitoring BCT' group. Over half (47/73) of the changes of parent class reflected the BCT being moved to a different higher-level group.The revised version of the BCTO following expert review had 247 BCTs arranged into 20 higher-level groups organised over five hierarchical levels (see https://osf.io/escjk)(West et al., 2020).
Step 5: Inter-rater reliability of annotations using the BCTO Inter-rater reliability from the 50 papers annotated by those familiar with the ontology was a=0.82 (see https://osf.io/7nqvb) and a=0.79 (see https://osf.io/u7dxs(West et al., 2020) for the 50 papers annotated by researchers unfamiliar with the ontology (Hayes & Krippendorff, 2007).These are considered good levels of inter-rater reliability.
Final revisions were made to the BCTO based on: a) issues raised by the annotators, b) revision of the BCTO by the research team on clarity and consistency of labels and definitions of the BCTs, and c) additional suggestions from behavioural scientists on BCTs that were missing or BCTs that still required more clarity.The following changes were made: • 34 BCTs were added.For example, two child classes were created for 'inform about health consequences BCT', which were 'inform about positive health consequences BCT' and 'inform about negative health consequences BCT' • The labels of 73 BCTs were updated.For example, 'advise how to perform behaviour BCT' was changed to 'guide how to perform behaviour BCT' • The definitions of 16 BCTs were updated.
Step 6: Computer-readable version of the BCTO and publication in online repositories The first release of the BCTO consisted of 281 BCTs hierarchically organised into 20 higher-level groups, with between one and 77 BCTs per higher-level group.The 20 higherlevel BCT groups, their definitions and number of BCTs per group are shown in Table 2.An excerpt from BCTO showing the BCTs belonging to the higher-level group "goal directed BCT" is shown in Table 3.To facilitate visualisation,

Advise how to change emotions BCT
A BCT that suggests a method to alter emotions.20

Restructure the environment BCT
A behaviour change technique that alters the environment in which the behaviour is, or would have been, performed in a way that facilitates or impedes the behaviour.
12 Prompt focus on self-identity BCT A BCT that prompts the person to focus on their mental representation of themself. 5

Behavioural consequence BCT
A behaviour change technique that alters the consequences or promised consequences for a behaviour.

Outcome consequence BCT
A behaviour change technique that alters the consequences or promised consequences for an outcome that results from performing or not performing a behaviour.

35
Note.The number of BCTs in each of group should not be considered final.The BCTO will continue to evolve and grow (e.g.new BCTs or new groups) as a result of user feedback and new available evidence.
Figure 3 shows the hierarchical relationships between entities in the "goal directed BCT" group, using the dedicated online tool, BCIOVisualise.
A downloadable version of the BCTO is available from GitHub (https://github.com/HumanBehaviourChangeProject/ontologies/tree/master/BehaviourChangeTechniques/inputs) and it can be browsed in the dedicated BCIOSearch tool and the Ontology Lookup Service.The hierarchical structure, Uniform Resource Identifiers (URI), labels and definitions for all entities are described in https://osf.io/ya74q(West et al., 2020).The ontology is accompanied by an annotation guidance manual on how to annotate entities in behaviour change intervention reports (https://osf.io/mwv2c(West et al., 2020)).A <goal directed BCT> that advises the person to find ways to remind themselves of their behavioural goal.
advise to keep outcome goal in mind BCT BCIO:007142 A <goal directed BCT> that advises the person to find ways to remind themselves of their goal in terms of a desired outcome of performing the behaviour.
Note.The BCTO forms part of the larger BCIO, therefore all entity URIs hold the form BCIO:XXXXXX.The BCTO will continue to evolve and grow as a result of user feedback and new available evidence, so it is likely that new BCTs will be added to this group or definitions will be updated.Most BCT definitions have been written in such a way that they can apply a) when the BCT is delivered by someone other than the person whose behaviour is being targeted, b) when the BCT is delivered without a person acting as the intervention source (e.g., in digital or print interventions), or c) when a BCT is self-enacted or initiated as part of a behaviour change attempt.
The exception to this rule is when the involvement of an intervention source as well as the person changing the target behaviour is inherent to the nature of the BCT.Examples of this include 'agree on behavioural goal BCT' and 'agree on outcome goal BCT', as an agreement needs to be made with another, ' create behavioural contract BCT' as part of this BCT is having the contract witnessed by another and ' observation of behaviour by another without feedback BCT'.In such cases, the definitions of the BCTs make clear that another person has to be involved.

Difference between a new BCT and instances of the same BCT.
When reading authors' descriptions of their interventions or when classifying BCTs during intervention development or evaluation, one may notice what appears to be a new BCT.BCTs can be added to the ontology when it is confirmed that they can be defined in a manner that differentiates them from the BCTs already included in the ontology.In most cases, these apparently new BCTs turn out to be particular implementations or child classes of BCTs that are already in the ontology.
Examples or the former are different solutions that can be used to overcome barriers to a given behaviour, or variations of action planning.Different child classes of a BCT can be added to the ontology if their specific features are important.

Preparatory behaviours
Behaviours that are required for a target behaviour to be performed are not behaviour change techniques per se.For example, obtaining a prescription for a medicine to aid smoking cessation is a requirement for taking the medication.

Combinations of BCTs
Some BCTs are often delivered alongside other BCTs.However, for simplicity and to avoid making the BCTO larger, we have not included classes representing combinations of BCTs because it would be impossible to include all the combinations of BCTs that intervention developers might view as important.Some intervention classification systems that we reviewed included categories that are combinations of BCTs.For example, in social prescribing, "connect to social support" is an intervention strategy (Cunningham et al., 2022).Examination of the definition of "connect to social support" revealed that it was a combination of two BCTs (advise to seek social support BCT followed by arrange social support BCT).Therefore, we have not included "connect to support" as a BCT.However, given the importance of this strategy in social prescribing, we have noted in the elaboration of "arrange social support BCT" and "advise social support BCT" that when used together, they may be termed "connect to support".When it applies, logically defined classes of BCT combinations can be added.

Advantages of BCTO over BCTTv1
Ontological systems may be specified in various ways (such as controlled lists, thesauri, taxonomies, or formal representations in logic), and these lie on a continuum of semantic complexity (National Academies of Sciences, Engineering, and Medicine, 2022).Formal ontologies (e.g., BCTO) include strong semantics (specifying properties of entities and their relations in formal logic), whereas taxonomies (e.g.BCTTv1) include weak semantics (may only specify parent-child relations).

Discussion
This study has developed a logically structured Behaviour Change Technique Ontology for describing and classifying BCTs using a computer-readable common terminology.The first published version consists of 281 BCTs organised into 20 higher-level groups and five hierarchical levels.It is published on an open-source platform alongside tools for visualisation and searching (https://github.com/HumanBehaviourChangeProject/ontologies/tree/master/BehaviourChangeTechniques).
The BCTO is part of the Behaviour Change Intervention Ontology, currently made up of 11 ontologies: intervention delivery mode (Marques et al., 2021), source (Norris et al., 2021), schedule and dose (in preparation), style (Wright et al., 2023), human behaviour (in preparation), mechanisms of action (Schenk et al., 2023), engagement (in preparation), fidelity (in preparation), and contextual influences such as intervention setting (Norris et al., 2020) and target population (Michie et al., 2020).The BCTO allows one to represent interventions in their contexts in a comprehensive and structured way enabling the answering of complex questions along the lines of: This improved method of specifying behavioural intervention content overcomes some limitations of BCTTv1 but will continue to need updating and improving.
Key differences between the BCTO and the Behaviour Change Technique Taxonomy v1 (BCTTv1) The BCTO contains considerably more BCTs and classes than BCTTv1.It also has a deeper hierarchical structure.In BCTTv1, the BCTs were organised over two levels.In the BCTO, BCTs are organised in a five-level hierarchy.The BCTO's hierarchy provides a more logical organisation of classes.For example, in the BCTTv1, "social support (unspecified)" and "social support (emotional)" are on the same level, even though emotional social support is a particular type of general social support.In the BCTO, "advise to seek emotional support BCT" is a child class of the parent class "advise to seek support BCT."The position of a BCT in the hierarchy reflects the organisation of the BCTO according to ontological principles.BCTs at the deeper levels of the hierarchy are more granular and specific than those at higher levels of the hierarchy.
The vast majority of BCTs from BCTTv1 can be mapped to one or more BCT in the BCTO (see https://osf.io/r7cux;West et al., 2020).In BCTTv1, the labels of groups were selected based on the group's content and where applicable, the frequency of words in labels provided by participants (Michie et al., 2013).BCTO labels reflect good practice in writing labels for ontology classes (Michie et al., 2019).Some BCTTv1 groups were labelled according to the hypothesised mechanism of action of the BCTs in that group (e.g."shaping knowledge" or "self-belief").In the BCTO, BCTs or their classes were not generally defined in terms of their potential mechanisms of action, because many BCTs can have more than one mechanism of action, depending on context, how the BCT is delivered, or which other BCTs are delivered at the same time.An Ontology of Mechanisms of Action (Schenk et al., 2023), also part of the Behaviour Change Intervention Ontology, can be used in conjunction with BCTO to describe both intervention content in terms of BCTs and their hypothesised mechanisms of action.The mapping from BCTTv1 to the BCTO (https://osf.io/r7cux;West et al., 2020), will be useful to those wishing to link up information classified by BCTTv1 with that classified by BCTO and those using BCTO whilst who also wish to use the Theory and Techniques Tool, an evidence-based method for linking BCTs with their hypothesised mechanisms of action (Johnston et al., 2021).We have also produced the 'reverse' mapping i.e. mapping BCTO to BCTTv1, so users can see all BCTs, including which are new and which can be mapped by back to BCTs in BCTTv1 (see https://osf.io/ru3q2).
Definitions of BCTs in BCTO conform to principles for writing "good" ontological definitions (Michie et al., 2019;Seppala et al., 2017).Each definition describes a BCT in terms Readers may ask why they should consider using BCTO over BCTTv1 (given the latter is widely used) and there are many reasons for this.Feedback from multiple sources (see Corker et al., 2023 for details) on BCTTv1showed: (i) the need to add more BCTs (more than 93 BCTs exist), (ii) the cluster labels from BCTTv1 were undefined and difficult to understand, and (iii) the taxonomical structure of BCTTv1 was based on cluster analyses, therefore not logical and did not allow the addition of BCTs.The BCTO has several advantages including: (i) being more complete than BCTTv1, (ii) having a logical structure (BCTs can be added), (iii) more precise and clear groupings, labels and definitions, (iv) links to other aspects of an intervention scenario, such as mechanisms of action, and (v) being computer-readable, which can support the application of artificial intelligence and machine learning approaches in data extraction, evidence synthesis, and outcome prediction.
of its parent class plus the things that differentiate the BCT from its parent class.To fully understand the nature of a BCT from its definition, ontology users will need to check the definition of the parent class stated in the first part of the definition.The structure of BCTTv1 was derived from a cluster analysis of experts' groupings of BCTs.As a result, it was not possible to add new BCTs to a group without repeating the expert grouping task and statistical analyses.Additionally, for some groups, there was no clear unifying feature of the BCTs in the group.The logical structure of BCTO overcomes these problems by having an explicit basis for the inclusion of BCTs within a group.Moreover, because the BCTO has a logically defined structure, new BCTs that are identified can be added to the ontology where they fit best, based on their ontological definitions.Future changes in the ontology can be recorded, along with explicit reasons for the changes.

Strengths and limitations
A strength of this study is the use of a standardised, tried and tested method for ontology development created within the Human Behaviour-Change Project (Wright et al., 2020) that reflects the OBO Foundry principles of good practice in ontology development (Norris et al., 2019;Smith et al., 2007).As a result, the BCTO fares well against several ontology evaluation criteria (Vrandečić, 2009).The BCTO's completeness, or how well it covers the domain of interest, was tackled by extending BCTTv1 and using scoping terms from other BCT classification systems.Completeness was also checked as part of the expert review, with experts asked if they thought any BCTs were missing from the ontology.The ontology's accuracy, in terms of how well it accords with experts' knowledge, was addressed by having the ontology developed by a team of researchers with considerable expertise in behaviour change interventions and by subjecting the ontology to expert review.The ontology's clarity, in other words whether it communicates the intended meaning of the defined classes, was examined through inter-rater reliability testing, which assesses whether independent annotators can agree on what constitutes an example of a BCT, using the definitions in the ontology.
A strength of the development of the BCTO is the use of international expert feedback in revising the ontology, a practice which has been uncommon in ontology development (Norris et al., 2019).Involving a range of experts provides a variety of perspectives on the ontology which is necessary to build consensus around definitions.Participating experts were recruited via social media and newsletter dissemination by the UCL Centre for Behaviour Change and by the Human Behaviour-Change Project.Ontology experts were recruited by the team's ontology expert.While the ontology benefits from the incorporation of international expert stakeholder feedback in the ontology development process, only half of invited behavioural science experts provided feedback on the ontology.With more time and resources, other means of engaging participants from under-represented parts of the world and disciplines could have been developed and are likely to have led to wider representation.Our aim is to disseminate the BCTO, and the Behaviour Change Intervention Ontology more generally, as widely as possible, to engage users and encourage feedback as the ontology is used in a variety of settings and for a variety of purposes.
The status and future of the BCTO Ontologies should be maintained and updated according to new evidence about entities and relationships (Arp et al., 2015;He et al., 2018).The development and maintenance of an ontology is an iterative process; no ontology is ever 'finished'.This is the first published version of the BCTO and it will continue to evolve.As with other ontologies produced as part of the Human Behaviour-Change Project (Michie et al., 2020), the BCTO will be refined through application and feedback from users via GitHub (https://github.com/HumanBehaviourChangeProject/ontologies/issues).For instance, if users believe additional entities are needed in the ontology, or if a BCT definition should be revised, then they make these suggestions on GitHub and these will be reviewed by the ontology developers.Guidance on how to do this can be found on the project website (https://www.bciontology.org/contribute).
If the suggestions are appropriate (as discussed and decided among the research team or through further stakeholder feedback), the development team will make the changes.User suggestions and subsequent changes will be made on a periodic basis (e.g., every 6 months) and we will explicitly document responses to user suggestions and changes between released versions of the ontology.Normally, ontologies are maintained for at least three years from the time of acceptance into the OBO Foundry.Collaborations formed through the BSSO Foundry (https://www.bssofoundry.org/)-a foundry for ontologies in behavioural and social sciences -can also support the evolution and maintenance of the BCTO.This commitment to ongoing updates and revisions to the BCTO creates opportunities for feedback from a broad range of experts to enhance and elaborate the ontology.The scope of the BCTO and associated ontologies within the Behaviour Change Intervention Ontology was limited to a level of generality considered to be of wide interest and use; those focussing on specific parts in more detail will need to extend the ontology.
It is also important that users report the unique identifier (e.g.BCIO: XXXXXX) in their intervention descriptions, so it is clear as exactly to which BCT(s) were used.Whilst entity labels and definitions can change over time (e.g. based on user feedback or new scientific evidence), the unique identifier always remains the same.Furthermore, as the content of ontologies can change over time, we also recommend that ontology users report the publication date of the ontology version that they used in their work.
The BCTO is designed to be connected ("interoperable") not only with the other parts of the Behaviour Change Intervention Ontology but also with ontologies in other fields, for example, health care, neuroscience, mental functioning, research methods and biology.It achieves this through the use of the standard ontology representation language, OWL, community-agreed metadata standards, and a common framework for the upperlevel structure of ontologies, the "Basic Formal Ontology" (BFO;Arp et al., 2015;Grenon & Smith, 2004;Smith & Grenon, 2004).BFO's upper-level structure divides things that exist in the world into two overarching categories: "continuants", which are objects and spatial entities that continue to exist as the same individual over time, such as an intervention's geographical setting, and "occurrents", which are events or processes that unfold in time.All the BCTs in the BCTO are processes.The BCTO's interoperability with ontologies from related fields creates exciting potential for future cross-disciplinary working and data integration.
The BCTO provides an improved method of specifying behavioural intervention content that overcomes identified limitations of the Behaviour Change Techniques Taxonomy v1; it will continue to need updating and improving.In the future, updated versions of the BCT Ontology will be released via GitHub [https://github.com/HumanBehaviourChangeProject/ontologies/tree/master/BehaviourChangeTechniques]and all updates will be available in the BCIOSearch and OLS tools as well as via the Behavioural and Social Sciences Ontology Foundry repository.We recommend that prospective users of the ontology check these online resources to ensure they have the most recent version of the ontology.

Conclusion
The BCT Ontology provides a common terminology and comprehensive structure for describing and classifying BCTs that can enable more efficient evidence accumulation and synthesis about 'what works' in behaviour change interventions across scientific disciplines and behavioural domains.This improved method of specifying behavioural intervention content extends and improves the Behaviour Change Techniques Taxonomy v1 but will continue to need updating and improving in an ongoing and collaborative process.The ontology is being published on an open-source platform alongside tools for visualisation and searching alongside other ontologies that form the Behaviour Change Intervention Ontology, providing a foundation on which future research on behaviour change can build on.

Consent
All participants provided their informed consent to participate in the study.The consent was obtained electronically through Qualtrics, the platform used for the survey.The participants indicated their consent by ticking a box.This consent process was in the ethics approval.
Thank you for asking me to review this work, I enjoyed reading it.There is a great need for such an ontology for both researchers and practitioners.This work potentially makes a behaviour change approach more accessible to non-specialist/non-psychologist researchers and health care practitioners.
The methods demonstrate a robust process and these were clear to me.I wondered why only four out of 17 invited behavioural scientist stakeholders participated and the in step 4 and the implications of this.Particularly as this small number of participants had so many comments.I wondered if a larger group may have been wise.That said, the definitions appeared to be effective enough for the assessors in step 5. From the discussion I understood the purpose, but I could not quite grasp the output of the machine-readable stage.I look forward to subsequent publications of this valuable work.

Is the study design appropriate and is the work technically sound? Yes
Are sufficient details of methods and analysis provided to allow replication by others?

If applicable, is the statistical analysis and its interpretation appropriate? Yes
Are all the source data underlying the results available to ensure full reproducibility?Yes

Are the conclusions drawn adequately supported by the results? Yes
Competing Interests: No competing interests were disclosed.
Reviewer Expertise: Implementation Science I confirm that I have read this submission and believe that I have an appropriate level of expertise to confirm that it is of an acceptable scientific standard.

Introduction:
In the 5th paragraph it would be helpful to make the differences between taxonomy and ontology a bit clearer (in my understanding mainly hierarchy versus network structure).
Perhaps it should also be made clear first that this is a computer science term related to Artificial Intelligence and its application to texts.This would help to clarify some formal requirements and rules of ontologies.

Last paragraph of introduction:
for clarity please add that the feedback comments were on BCTTv1.

○
When first mentioning the BCI ontology it may be useful to emphasise "Intervention"(for example writing it in italics) to distinguish it from BCTO.Since the work describes a complex development process related to many other pieces of work from the Human Behaviour Change project it may sometimes be hard for readers not familiar with the project to distinguish between all components, especially as they share integral constituents within their names and many similar abbreviations are used.

○
In this context, I would appreciate it if a graphical overview of the project could be given, in which the parts named in the text (HBCP, BCIO, and the different ontologies on the same level as BCTO) are illustrated and shown in their connections, so that it is easier to classify them when reading.Such a diagram could replace the current Figure 1, as this essentially contains the headings to the design steps from the text and thus no additional information is given.

Objective:
The aim of the work is clearly stated.Again, for readers who are not very familiar with AI tools, it might be useful to briefly explain what computer-or machine-readable means exactly, what requirements the ontology must fulfil for this, and to what extent the BCTO is intended to be used by humans and/or computer applications.

Glossary:
My impression with the glossary was that some terms defined here are rather less important for the BCTO itself than for the overall project.I wonder whether more technical entries on e.g.GitHub, versioning, OBO Foundry principles or ROBOT should actually be listed here.The latter, for example, occurs quite far back in the text and could simply be explained briefly by a half-sentence there.

○
On the one hand, these are certainly terms that many readers will not know and therefore it can be useful to explain them in the glossary.On the other hand, they do not always seem necessary to me for understanding the ontology and its structure (e.g. the specific software or websites).It could also make sense to divide the glossary into more content-related ontology principles and terms versus more technical terms related to the tools used.

○
The terms "class" and "entity" are defined differently but used interchangeably.This is somewhat confusing.For example, the definition of "entity" also refers to "classes" in addition to processes and attributes.It would improve readability if, at least in the text, a uniform term were predominantly used.

○
In the entry on "GitHub", it could be added that "code" refers to computer code.

○
The entry on OWL talks about "things".I think one could also speak of classes or entities here, so as not to use yet another term?
○ Not all of the terms explained can be found in the text.For example, "OBO Foundry Principles" are not mentioned in the text.These principles could also be added to the main entry on OBO."Issue tracker" is highlighted in the text but not in the glossary.

Design:
Step 4: Under this step it is first described that all experts should review all BCT groups.Then it says "Participants were presented with the label of a single higher-level group from the BCTO and all the BCTs within that group."-Should this mean "one group at a time" or "one after the other"?Otherwise it seems contradictory.

Results:
I really much appreciate the extensive documentation of each step through additional online files linked in the text.

Discussion:
The reference to a document with the mapping between BCTTv1 and BCTO annotations is very helpful.Overall, it seems important to me that intervention descriptions made on the basis of older versions can be assigned as clearly as possible to the current BCTO version, because otherwise part of the problem remains with different terms used for the same techniques.

○
It would be helpful for potential users to describe more clearly how an update of the annotations for translations into the latest BCTO version can look in the future.This also seems important to me precisely because the BCTO can develop further and thus change.
○ Future prospect: "Future changes in the ontology can be recorded, along with explicit reasons for the changes."Keeping the ontology up to date, checking feedback and changing or adding entries if necessary will require a lot of work.Also, because different ontologies refer to each other.But it seems to me an important prerequisite to benefit from the ontology system in the best possible way, because some problems may only emerge over time with practical application.
"...will continue to need updating and improving in an ongoing and collaborative process." Here it seems helpful to me to communicate current considerations in more detail.It is possible that procedures will also result from the principles of the open source network, but a few aspects do not seem clear enough to me and maybe others: ○ Should the project continue as an open science project in which all interested users can generally participate?Will a qualified core team continue to be responsible for the adaptations?Above all: how will it be decided whether entries should be changed, according to which criteria and who can make changes?After all, an ontology that changes too frequently can no longer be used uniformly and be consensual.

○
The link to the BCIO main page contains a typo and therefore does not work: Please remove superfluous "o" from www.bcioontology.org/training

Is the work clearly and accurately presented and does it cite the current literature? Yes
Is the study design appropriate and is the work technically sound?Yes

Are sufficient details of methods and analysis provided to allow replication by others? Yes
If applicable, is the statistical analysis and its interpretation appropriate?

Not applicable
Are all the source data underlying the results available to ensure full reproducibility?Yes

Are the conclusions drawn adequately supported by the results? Yes
Competing Interests: No competing interests were disclosed.
Reviewer Expertise: Health Psychology, Public Health, Behaviour Change I confirm that I have read this submission and believe that I have an appropriate level of expertise to confirm that it is of an acceptable scientific standard.
information unambiguously and effortlessly.According to the Open Biological and Biomedical Ontology (OBO) Foundry principles of good practice for developing ontologies, ontologies must be made available in a common formal language in an accepted concrete syntax (Principle 2), which is typically the Web Ontology Language (OWL)."We have also added to the last paragraph of the introduction a description of how the BCTO (and wider BCIO) is intended to be used by both humans and computers: "The BCIO is intended to be used by both humans and computers.Humans can use the ontology to describe and report interventions, synthesise evidence (e.g. through literature annotation), and develop interventions, to name a few possible use cases.In computer-readable form, the ontology is represented in OWL which can be parsed and used in several different downstream applications.For example, we can harness the power of computation in terms of Machine Learning and Artificial Intelligence to automate the extraction of information from reports (West et al. 2023), and build systems that can predict behaviour in novel scenarios (Hastings et al. 2023).Ontologies provide an interface between computers and humans, that is, computers can better support human tasks that require understanding of a domain, coupled with integration of content at scale and speed."

Glossary:
My impression with the glossary was that some terms defined here are rather less important for the BCTO itself than for the overall project.I wonder whether more technical entries on e.g.GitHub, versioning, OBO Foundry principles or ROBOT should actually be listed here.The latter, for example, occurs quite far back in the text and could simply be explained briefly by a half-sentence there.

○
Response: The glossary is intended to include low frequency terms or terms that may be unfamiliar to a behavioural and social science audience (especially terms relating to ontologies).Additionally, previous feedback from users have suggested that these technical terms were necessarily to include, as there was unfamiliarity with these terms.Therefore, we have decided to keep these entries in.
On the one hand, these are certainly terms that many readers will not know and therefore it can be useful to explain them in the glossary.On the other hand, they do not always seem necessary to me for understanding the ontology and its structure (e.g. the specific software or websites).It could also make sense to divide the glossary into more content-related ontology principles and terms versus more technical terms related to the tools used.

○
Response: To make the glossary most user-friendly, we have decided to have one glossary that orders the terms in alphabetical form.As it is a somewhat challenging task to separate terms into 'ontology principles', 'tools', and so on, we will not divide the glossary.The terms "class" and "entity" are defined differently but used interchangeably.This is somewhat confusing.For example, the definition of "entity" also refers to "classes" in addition to processes and attributes.It would improve readability if, at least in the text, a uniform term were predominantly used.Response: Classes in ontologies represent types of entities in the world.The terms 'entity' and 'class' are often used interchangeably in the literature to refer to entities represented in an ontology.To avoid confusion, we have amended the fifth paragraph of the introduction to be more precise.In the entry on "GitHub", it could be added that "code" refers to computer code.Response: We have now added this.The entry on OWL talks about "things".I think one could also speak of classes or entities here, so as not to use yet another term?
○ Response: We have now changed "things" to "entities".Not all of the terms explained can be found in the text.For example, "OBO Foundry ○ Principles" are not mentioned in the text.These principles could also be added to the main entry on OBO."Issue tracker" is highlighted in the text but not in the glossary.Response: We have added a sentence in the Discussion that include "OBO Foundry Principles" in the text, and have added "Issue tracker" in the glossary table.Design: Step 4: Under this step it is first described that all experts should review all BCT groups.Then it says "Participants were presented with the label of a single higher-level group from the BCTO and all the BCTs within that group."-Should this mean "one group at a time" or "one after the other"?Otherwise it seems contradictory.Response: Participants were asked to review BCT groups (and the BCTs within that group) one at a time.We have clarified this in the text in Step 4.

Results:
I really much appreciate the extensive documentation of each step through additional online files linked in the text.Response: Thank you for your comments.

Discussion:
The reference to a document with the mapping between BCTTv1 and BCTO annotations is very helpful.Overall, it seems important to me that intervention descriptions made on the basis of older versions can be assigned as clearly as possible to the current BCTO version, because otherwise part of the problem remains with different terms used for the same techniques.

○
Response: We agree that it is important that authors reporting intervention descriptions based on BCTTv1 might also consider mapping BCTs (from BCTTv1) to BCTO, using the mapping document.We have added a few sentences in the Discussion ("The status and future of the BCTO" section) to address the issue of reporting which version of the ontology authors use: "It is also important that users report the unique identifier (e.g.BCIO: XXXXXX) in their intervention descriptions, so it is clear as exactly to which BCT(s) were used.Whilst entity labels and definitions can change over time (e.g. based on user feedback or new scientific evidence), the unique identifier always remains the same.Furthermore, as the content of ontologies can change over time, we also recommend that ontology users report the publication date of the ontology version that they used to their work." It would be helpful for potential users to describe more clearly how an update of the annotations for translations into the latest BCTO version can look in the future.This also seems important to me precisely because the BCTO can develop further and thus change.

○
Response: Thank you for raising this important issue.We have added to the section ("The status and future of the BCTO") to address this issue: "Ontologies should be maintained and updated according to new evidence about entities and relationships ( Arp et al., 2015;He et al., 2018).The development and maintenance of an ontology is an iterative process; no ontology is ever 'finished'.This is the first published version of the BCTO and it will continue to evolve.As with other ontologies produced as part of the Human Behaviour-Change Project ( Michie et al., 2020), the BCTO will be refined through application and feedback from users via GitHub ( https://github.com/HumanBehaviourChangeProject/ontologies/issues).For instance, if users believe additional entities are needed in the ontology, or if a BCT definition should be revised, then they make these suggestions on GitHub and these will be reviewed by the ontology developers.Guidance on how to do this can be found on the project website ( https://www.bciontology.org/contribute).If the suggestions are appropriate (as discussed and decided among the research team or through further stakeholder feedback), the development team will make the changes.User suggestions and subsequent changes will be made on a periodic basis (e.g., every 6 months) and we will explicitly document changes between released versions of the ontology.Normally, ontologies are maintained for at least three years from the time of acceptance into the OBO Foundry.Collaborations formed through the BSSO Foundry ( https://www.bssofoundry.org/)-a foundry for ontologies in behavioural and social sciences -can also support the evolution and maintenance of the BCTO."Future prospect: "Future changes in the ontology can be recorded, along with explicit reasons for the changes."Keeping the ontology up to date, checking feedback and changing or adding entries if necessary will require a lot of work.Also, because different ontologies refer to each other.But it seems to me an important prerequisite to benefit from the ontology system in the best possible way, because some problems may only emerge over time with practical application.
"...will continue to need updating and improving in an ongoing and collaborative process." Here it seems helpful to me to communicate current considerations in more detail.It is possible that procedures will also result from the principles of the open source network, but a few aspects do not seem clear enough to me and maybe others: ○ Should the project continue as an open science project in which all interested users can generally participate?Will a qualified core team continue to be responsible for the adaptations?Above all: how will it be decided whether entries should be changed, according to which criteria and who can make changes?After all, an ontology that changes too frequently can no longer be used uniformly and be consensual.

○
Response: The ontology indeed needs to reflect changes in scientific consensus to remain accurate over time.Updating and maintaining the ontology is a key principle in the OBO Foundry best practice for developing ontologies.The ontology will continue to evolve as an "open science project" in the sense that anyone can suggest changes through the "issue tracker" portal of Github.At least one member of the development team will be responsible for reviewing these user suggestions, and these will be discussed and decided among the research team or through further stakeholder feedback.Current team members are committed to doing this for at least the next six years.We will also ensure that the responsibility is handed over to others with the necessary expertise after this time period.As you rightly mentioned, ontology maintenance requires a lot of work, therefore, reviews and changes by the development team would occur on a periodic basis (e.g.every six months during the initial years of use, then less frequently such as yearly).Normally, ontologies are maintained for at least three years from the time of acceptance into the OBO Foundry.Changes will only be made if absolutely necessary -there must be good reason for these changes (e.g.new scientific evidence coming to light; a majority consensus from multiple users).
The link to the BCIO main page contains a typo and therefore does not work: Please remove superfluous "o" from www.bcioontology.org/trainingThe manuscript describes in detail a large, rigorously conducted, body of research that was used to achieve the current BCT Ontology (BCTO).Supplementary materials are provided to evidence all parts of the development process.There is also a process to update the BCTO that will futureproof this work.This BCTO will certainly benefit the science of behaviour change (and I'm excited to start using it!).
I do have a few minor comments: Response: The three levels referred to an initial version of the BCTO, that was then expanded to 5 hierarchical levels.Levels correspond to parent and child classes, and this structure is explained in the results section.
Examples throughout would be really useful.Where provided they make things a lot clearer without having to access the supplementary materials I commend the authors on their rigorous and comprehensive approach to improving behaviour change research.This article provides a helpful overview of the research journey from the BCTT to the BCTO.I was generally able to follow along with the authors as they described gathering feedback on the BCTT; updating entity labels and descriptions; creating hierarchical relationships between entities to define the ontology; soliciting expert feedback; validating coding with interrater reliability; and publication.There were a couple of occasions where the balance of detail and clarity made it hard to understand exactly what happened at each step; I point these out below.
Overall I thought this article describes an extremely strong contribution to behaviour change research.I have several minor recommendations, which are offered as collegial suggestions only; if the authors have a different preference I encourage them to not implement them if they feel it will not improve the value of the paper.
1. Figures: Figure 1 and Figure 2 could be combined, because they describe the same process.
Figure 2 was somewhat inconsistent in its presentation; e.g., sometimes the number of BCTs at the end of each step was mentioned, and sometimes it was omitted.Figure 2 had some minor spelling errors (e.g., emended instead of amended).
2. Examples: when authors provided examples, it was much easier to understand each element of the process from initial BCTT to final BCTO.However, the inclusion of examples was inconsistent, e.g., for Step 2, an example was provided where 19 BCTs were split, but no other examples were provided for the other actions at this step.It did seem as though the goal directed BCTs was used throughout the process as an example where they were relevant, including in Table 3, but when the goal directed BCTs were not relevant no example was provided.
3. Table 1 -Glossary: I appreciate the inclusion of a glossary to assist readers in unfamiliar terms, but I was confused by the terms that were included (and omitted) from the glossary.For example, the elements that comprise an ontology were included (e.g., entity, class, relationship, ontology), but also common software tools/platforms (e.g., EPPI-Reviewer and GitHub).I didn't understand why both 'class' and 'parent class' were included.'Ontology' was defined, but 'taxonomy' was omitted.Finally, the term 'BCI content' was defined, but the acronym BCI was not spelled out in the definition.I suggest a review of the glossary to ensure it is as useful as possible for readers.3 -Excerpt from the BCTO: I thought this excerpt helped to communicate the hierarchical relationship between the entities in the ontology.However, I could imagine it being even more useful if presented as a hierarchical figure, such as a tree diagram.This would more explicitly show the relationships between entities (e.g., through branches on the tree).A separate comment for Table 3 is that the inclusion of the BCIO codes (e.g., BCIO:007001) is somewhat confusing because that the table purports to be an excerpt of the BCTO, not the BCIO.In the Discussion, I do see that the BCTO will form part of the Behaviour Change Intervention Ontology (BCIO), so all BCTO entities would therefore have a BCIO code.However, it may assist readers include a clarifying note for the table, or omit the BCIO code.

Table
5. Clarifications for non-expert readers: the use of Box 1 to answer or clarify some questions that arose during the development process for the BCTO was very helpful.Two other clarifications -already described in text but useful to highlight -could clarify questions for interested, nonexpert readers.The first clarification is about the difference between a taxonomy and an ontology, and why an ontology was needed even though the BCTT already exists and is widely used.This is already addressed in the introduction, but making this comparison more prominent could help a reader who knows about the BCTT, but not the BCTO and its advantages.

○
The second clarification is about the difference between a behaviour change technique and a behaviour change intervention.This is mentioned several times in the introduction (e.g., "Behaviour change techniques are defined as…" and later "ontologies… can link BCTs to other intervention features such as their delivery…") but again, making this comparison more prominent could help a reader understand how BCTs 'fit' within an intervention, and how the BCTO fits with the larger BCIO.Reviewer Expertise: Behaviour science I confirm that I have read this submission and believe that I have an appropriate level of expertise to confirm that it is of an acceptable scientific standard.

Table 1 -Glossary:
I appreciate the inclusion of a glossary to assist readers in unfamiliar terms, but I was confused by the terms that were included (and omitted) from the glossary.For example, the elements that comprise an ontology were included (e.g., entity, class, relationship, ontology), but also common software tools/platforms (e.g., EPPI-Reviewer and GitHub).I didn't understand why both 'class' and 'parent class' were included.'Ontology' was defined, but 'taxonomy' was omitted.Finally, the term 'BCI content' was defined, but the acronym BCI was not spelled out in the definition.I suggest a review of the glossary to ensure it is as useful as possible for readers.Response: The glossary was intended to include low frequency terms or terms that may be unfamiliar to a behavioural and social science audience (e.g.terms relating to ontologies).Both 'class' and 'parent class' are included as they refer different things.'Classes' in ontologies represent types of entities ('things') in the world.They can be arranged hierarchically by the specification of parent and child classes.A 'parent class' is a class within an ontology that is hierarchically related to one or more child classes such that all members of the child class are also members of the parent class, and all properties of the parent class are also properties of the child class.We have made the following amendments to the glossary: Spelling out and defining 'BCI': A behaviour change intervention (BCI) is defined as "An intervention that has the aim of influencing human behaviour."(Michie et al. 2021) ○ Defining taxonomy: A taxonomy is defined as "A hierarchy consisting of terms denoting types (entities) which are linked by subtype relations."(Arp et al. 2015) ○ 4. Table 3 -Excerpt from the BCTO: I thought this excerpt helped to communicate the hierarchical relationship between the entities in the ontology.However, I could imagine it being even more useful if presented as a hierarchical figure, such as a tree diagram.This would more explicitly show the relationships between entities (e.g., through branches on the tree).A separate comment for Table 3 is that the inclusion of the BCIO codes (e.g., BCIO:007001) is somewhat confusing because that the table purports to be an excerpt of the BCTO, not the BCIO.In the Discussion, I do see that the BCTO will form part of the Behaviour Change Intervention Ontology (BCIO), so all BCTO entities would therefore have a BCIO code.However, it may assist readers include a clarifying note for the table, or omit the BCIO code.Response: We have now included a diagram/figure showing the relationships between entities for the 'goal directed BCT' group.This was generated via the dedicated online, open-access tool, BCIOVisualise (https://bciovis.hbcptools.org/).We hope this is helpful for readers to visualise the hierarchical relationship between entities.As the visualisation of large, complex ontologies can be challenging, we have only included a snapshot (i.e. of the 'goal directed BCT' group), rather than the entire BCTO.For the BCIO unique IDs (e.g., BCIO:007001), you are correct that the BCTO forms part of the larger BCIO, which is why all entity URIs within BCIO (including BCTs, mode of delivery, mechanism of action ontologies etc.) begin with "BCIO".To avoid confusion, we have added the following note for the table: "The BCTO forms part of the larger BCIO, therefore all entity URIs hold the form BCIO:XXXXXX" 5. Clarifications for non-expert readers: the use of Box 1 to answer or clarify some questions that arose during the development process for the BCTO was very helpful.Two other clarifications -already described in text but useful to highlight -could clarify questions for interested, non-expert readers.The first clarification is about the difference between a taxonomy and an ontology,

Figure 2 .
Figure 2. Summary of results of the steps for the BCTO development.

4
Prompt thinking related to successful performance BCTA behaviour change technique that prompts thinking relating to successful performance of a behaviour.6Changethe body BCTA behaviour change technique that alters the structure or functioning of the person's body.1Promotepharmacological support BCTA behaviour change technique promoting medicines or other drugs.3 Some important general observations and clarifications about the BCTO that emerged during the development process are outlined in Box 1.They concern delivery of BCTs by self or others, differences between a new BCT and instances of the same BCT, preparatory BCTs and combinations of BCTs.

Figure 3 .
Figure 3. Snapshot of the hierarchical structure of the "goal directed BCT" group.

○
Response: Thank you for noticing the typo.It has now been corrected.Competing Interests: No competing interests were disclosed.Reviewer Report 15 August 2023 https://doi.org/10.21956/wellcomeopenres.21449.r63730© 2023 Epton T. This is an open access peer review report distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.Tracy Epton 1 Manchester Centre for Health Psychology, Division of Psychology and Mental Health, School of Health Sciences, Faculty of Biological, Medical and Health Sciences, The University of Manchester, Manchester, England, UK 2 Manchester Centre for Health Psychology, Division of Psychology and Mental Health, School of Health Sciences, Faculty of Biological, Medical and Health Sciences, The University of Manchester, Manchester, England, UK The manuscript describes the development of a Behaviour Change Technique Ontology building on their former Behaviour Change Technique Taxonomy.The work described moves beyond the taxonomy of Behaviour Change Techniques (BCT) by refining existing and adding further BCTs, and situating the BCTs in an ontological framework to clarify the BCT properties and the relationships between the BCTs.

○
Response: We have now provided examples in steps 2-5, as suggested.We hope these added examples are helpful to show how BCTs were transformed from BCTTv1 to BCTO.Competing Interests: No competing interests were disclosed.Reviewer Report 07 August 2023 https://doi.org/10.21956/wellcomeopenres.21449.r63998© 2023 Saeri A. This is an open access peer review report distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.Alexander K Saeri 1 BehaviourWorks Australia, Monash Sustainable Development Institute, Monash University, Clayton, Victoria, Australia 2 BehaviourWorks Australia, Monash Sustainable Development Institute, Monash University, Clayton, Victoria, Australia The authors describe the research process to construct the Behaviour Change Technique Ontology (BCTO), an evolution of the widely used Behaviour Change Technique Taxonomy (BCTT), which describes more than 200 techniques that can change human behaviour, and the relationships between them.The article summarises and integrates six major sequential research activities described in detail in previously published work, which lead to an evidence and expertinformed ontology that can be used by researchers and machines to precisely define the content of behaviour change interventions, with an aim to improving future behaviour change research.

○
Is the work clearly and accurately presented and does it cite the current literature?YesIs the study design appropriate and is the work technically sound?YesAre sufficient details of methods and analysis provided to allow replication by others?Yes If applicable, is the statistical analysis and its interpretation appropriate?Not applicable Are all the source data underlying the results available to ensure full reproducibility?YesAre the conclusions drawn adequately supported by the results?YesCompeting Interests: No competing interests were disclosed.

Table 2 . Definitions of the 20 higher-level groups in the final BCT Ontology, and number of BCTs in each group (after step 5).
7Associative learning BCTA behaviour change technique that involves repeated pairing of a stimulus with another stimulus or with a behavioural outcome.15AdvisespecificbehaviourBCTAbehaviour change technique that advises the person to perform a behaviour in a particular way to help change the target behaviour.9Managementalprocesses BCTA behaviour change technique that advises how to manage mental processes to facilitate the target behaviour.

Table 3 . Excerpt from the BCTO presenting the 'goal directed BCT' group. Level 1 label Level 2 label Level 3 label Level 4 label
(Michie et al., 2020;aviour change interventions: What works, compared with what, for what behaviours, how well, for how long, with whom, in what setting, and why?" Answering variants of this question requires large quantities of data and sophisticated analyses; it requires automation and the application of Artificial Intelligence to identify relevant studies, extract relevant information and organise it within an ontology so that predictions can be made, drawing on the full range of intervention and contextual features.This was the aim of, and was mostly achieved, by the Human Behaviour-Change Project(Michie et al., 2020;https://www.humanbehaviourchange.org); it represents a step-change in the potential to accumulate evidence to address complex behavioural questions, thereby improving theories of behaviour change, the development of more effective interventions and the science of behaviour change more generally.

Table 1
Has some missing terms e.g., taxonomy, BCT ○Has some acronyms not spelled out e.g., BCI, EPPI ○ Examples throughout would be really useful.Where provided they make things a lot clearer without having to access the supplementary materials ○ Is the

work clearly and accurately presented and does it cite the current literature? Yes Is the study design appropriate and is the work technically sound? Yes Are sufficient details of methods and analysis provided to allow replication by others? Yes If applicable, is the statistical analysis and its interpretation appropriate? Not applicable Are all the source data underlying the results available to ensure full reproducibility? Yes Are the conclusions drawn adequately supported by the results? Yes Competing Interests
: No competing interests were disclosed.Reviewer Expertise: Behaviour Change Interventions; Behaviour Change Techniques I confirm that I