Skip to navigation – Site map

HomeIssuesWords about #1PapersCogMemo: a standardized, structur...

Papers

CogMemo: a standardized, structured and formalized terminological repository on human memory

Frank Arnould

Abstracts

This article presents the principles underpinning the modelling of CogMemo, a bilingual (French and English) thesaurus for the cognitive psychology of human memory, based on Semantic Web technologies and standards. The aims of CogMemo are to gather, formalize, standardize, disambiguate and organize the terminology in this area, with particular focus on memory systems and processes, the phenomenology of memories, study methods, measurements, empirical effects, theories, models and memory disorders. Suggestions are made as to possible uses for CogMemo. The difficulties of constructing this terminological repository, and its current limitations and future development, are also discussed.

Top of page

Full text

Acknowledgments
Richard Dickinson and Kate Kean (English translation), Majid Khayari and Lisa Amram (comments on the manuscript), Enza Morale and Stéphane Schneider (Istex corpora on memory).

1. Introduction

1According to the philosopher of science Carl Hempel [1966: 85], “[s]cientific statements are typically formulated in special terms [...]. If those terms are to serve their purpose, their meanings will have to be so specified as to make sure that the resulting statements are properly testable and that they lend themselves to use in explanations, predictions, and retrodictions.”

2Standardizing terminology and nomenclature therefore appears an essential step in any work towards scientific progress (Uddin et al. [2023: 825]). Using a common language with agreed terms denoting concepts in a particular field guarantees enhanced communication within and between scientific communities. It facilitates sharing knowledge obtained through laboratory research and its transfer to the field. A standardized vocabulary also means various types of data (datasets, scientific articles, brain maps, etc.) can be described in an explicit and unambiguous way, thus facilitating and ensuring the data’s interoperability, re-use and integration. Standardized vocabularies thus help improve the interpretation, reproduction, replication and generalization of research results.

3It is not a new development that the psychology community should wish to work with a standardized terminology. At the VIth International Congress of Psychology in Geneva in 1909, Édouard Claparède (1873-1940) had already denounced a certain confusion in his discipline’s use of terms, stating that “[m]ost often this confusion is caused by the lack of a precise nomenclature; many of the differences we think to be doctrinal are in fact simply down to different uses of words.” (Claparède [1910: 467], my translation). The Swiss psychologist went on to opine that the time had come to standardize the vocabulary of psychology.

  • 1 The typographical conventions are as follows (according to L’Homme [2020]): terms are written in ca (...)

4And yet the status of terminology in psychology has changed little over the following century and the field of cognitive psychology of human memory and associated concepts is no exception to this situation. Researchers regularly report problems of polysemy (Aimé & Arnould [2021]; De Brigard [2023]; Tulving [2000]). For example, the Canadian psychologist Endel Tulving has listed no fewer than six different definitions of the term memory1, “(1) memory as a neurocognitive capacity to encode, store, and retrieve information; (2) memory as a hypothetical store in which information is held; (3) memory as the information in that store; (4) memory as a property of that information; (5) memory as a componential process of retrieval of that information; and (6) memory as an individual’s phenomenal awareness of remembering something.” (Tulving [2000: 36]). This is also the case for more specific terms like working memory (Adams et al. [2018]; Cowan [2017]; Logie et al. [2021]). Also, several terms may sometimes denote the same concept. For example, the term FALSE MEMORY is commonly used to represent an erroneous memory of a complete event or of the detail of an event. However, other expressions, like pseudo-memory or MEMORY illusion, can also be found in the literature.

5The thesaurus seems an interesting tool for starting to standardize and structure the terminology of the cognitive psychology of memory. Firstly, this type of terminological repository includes a controlled vocabulary, namely a collection of preferred terms that are prescribed to denote concepts, which means that a common homogeneous language can become possible. Synonyms are indicated as alternative labels (equivalence relations between terms). Secondly, a thesaurus uses hierarchical and associative relationships to organize concepts in a specialized field. These relationships help define the way entities in a domain are divided and organized among themselves and also serve to resolve certain ambiguities. Thirdly, the design of a thesaurus now benefits from standards (ISO 25964 standards; International Organization for Standardization [2011], [2013]) to provide support for terminology work. Knowledge organization systems like SKOS (Simple Knowledge Organization System) help to formalize the conceptual and lexical knowledge of a thesaurus, to represent it on the Semantic Web and to link it to other data. It also makes the thesaurus machine-readable – for example, for automatically indexing and annotating scientific articles.

6CogMemo is a bilingual (French and English) thesaurus for the cognitive psychology of human memory which is available in open access and constructed using these recommendations and technologies. The objective underpinning this repository is to use the SKOS model to carry out the standardization, disambiguation and formalization of terminology collected in this domain and associated fields. CogMemo focuses on memory systems and processes, memory measures, methods, models, theories and disorders along with the subjective aspects of memories (phenomenology).

7Scientific and terminological monitoring of published literature on cognitive approaches to memory in humans is the basis for populating CogMemo, which is complemented by terminology extraction and bibliometric tools working on corpora of targeted scientific articles. The thesaurus is managed using VocBench, a collaborative web application for creating and editing ontologies, thesauri and lexicons for the Semantic Web (Stallato et al. [2020]).

8The first version of CogMemo was put online in April 2018. On average, it is updated twice a year to make sure it is effectively maintained and continues to evolve. Currently 1,409 concepts are listed at the time of writing (version 2.4). The repository can be consulted online at Loterre – a SKOS terminology exposure platform2 – and can also be downloaded in PDF, SKOS / XML and CSV formats.

9The rest of this article will describe the general modelling principles for CogMemo using the SKOS model on the Semantic Web. Following this, suggestions are made of usage examples of the repository. Finally, the difficulties encountered during its design, its current limitations and future development prospects are discussed.

2. CogMemo: concepts, terms and the Semantic Web

10The SKOS model3 is a W3C (World Wide Web Consortium) recommendation. It is used to express the basic structure of thesauri, classification schemes and other controlled vocabularies on the Semantic Web4, “[a] set of standards and best practices for sharing data and the semantics of that data over the Web for use by applications” (DuCharme [2013: 501]).

11A thesaurus that is formalized in SKOS is primarily organized around concepts with terms used to denote them. This means that its mode of operation differs somewhat from that of a standard dictionary. For example, in the American Psychological Association’s Dictionary of Psychology, the entry MEMORY is followed by three definitions: “1. the ability to retain information or a representation of past experience, based on the mental processes of learning or encoding, retention across some interval of time, and retrieval or reactivation of the memory. 2. specific information or a specific past experience that is recalled. 3. the hypothesized part of the brain where traces of information and past experiences are stored5”. In the conceptual approach to terminology taken up by the SKOS model and the ISO-25964 standards, these definitions could potentially represent three different concepts and thus three thesaurus entries. However, the term MEMORY cannot be used to denote these three entries. This means that contrasting lexicalizations will need to be formulated to represent the different concepts.

12SKOS provides thesaurus designers with classes and properties to describe concepts, their lexicalizations, documentation (definitions, notes, etc.) and relationships. All CogMemo’s concepts are instances of the skos:Concept class. Each entry in the thesaurus (e.g. the concept « episodic memory ») is a resource with a unique permanent Uniform Resource Identifier (or URI6) to reference it on the internet and link it to other resources. Facts (for example, “episodic memory” has for broader concept declarative memory”) are represented using the RDF (Resource Description Framework) data model in the form of Subject – Predicate – Object triplets. The subject of a triplet corresponds to the described resource (the « episodic memory » concept, identified by its URI). The predicate is a property (has for broader concept), while the object is the value of this property which is another resource (in this example, the resource that identifies the « declarative memory » concept) or a literal (for example, for the lexicalization of a concept).

13However, the SKOS model does not enable all the knowledge represented in CogMemo to be expressed. To achieve this, the thesaurus introduces new properties or borrows them from other metadata vocabularies. These new properties will be presented subsequently.

2.1. Lexicalization of concepts

14Three SKOS properties formalize the lexicalization of concepts in CogMemo (see Table 1 for an example). The mandatory skos:prefLabel property advises which preferred term should be used to denote a concept. The set of preferred labels thus means CogMemo can be defined as a controlled vocabulary of human memory. Only one preferred label is authorized per language and it is selected according to the frequency of use of the terms in the literature. One preferred term cannot be used to denote different concepts. Disambiguation strategies have to be implemented to deal with such situations. For example, REcognition can refer, at least, to two concepts – the ability to discriminate between memorized and new items, and the task of doing so in itself. In CogMemo, the preferred terms recognition memory and recognition task respectively denote these two meanings.

Table 1. Example of the lexicalization of a concept in CogMemo using SKOS for the concept « false memory » in the form of RDF triplets (extract)

Concept (subject)

property (predicate)

litteral@lang (object)

http://data.loterre.fr/​ark:/67375/​P66-JQ343CBV-S

skos:prefLabel

“false memory”@en

http://data.loterre.fr/​ark:/67375/​P66-JQ343CBV-S

skos:prefLabel

“faux souvenir”@fr

http://data.loterre.fr/​ark:/67375/​P66-JQ343CBV-S

skos:altLabel

“memory illusion”@en

http://data.loterre.fr/​ark:/67375/​P66-JQ343CBV-S

skos:altLabel

“illusion mnésique”@fr

http://data.loterre.fr/​ark:/67375/​P66-JQ343CBV-S

skos:hiddenLabel

“false memories”@en

http://data.loterre.fr/​ark:/67375/​P66-JQ343CBV-S

skos:hiddenLabel

“faux souvenirs”@fr

15The skos:altLabel property indicates the synonyms of the preferred terms. Except in the case of established usages, abbreviations and acronyms are given as alternative labels for expanded expressions which themselves are chosen as preferred labels. The skos:hiddenLabel property enables lexicalizations to be introduced into a thesaurus which are not displayed online but can be used, for example, by a search engine or to enhance the automatic indexing and annotation of documents. In CogMemo, the hidden labels essentially represent lexicographical variants or the plurals of preferred and alternative lexicalizations. This is also a useful property for incorporating common spelling errors, although it should be noted that CogMemo does not yet take this usage into account.

Table 2. The number of preferred, alternative and hidden labels in English and French in CogMemo

Language

Preferred labels

Alternative labels

Hidden labels

English

1409

1407

883

French

1409

1271

869

2.2. The documentation of concepts

16Above all, a SKOS thesaurus is a conceptual artefact. The classical theory of concepts identifies concepts with definitions (Laurence & Margolis [1999]; Machery [2010]; Murphy [2002]). Therefore, the role of definitions in CogMemo is to delimit the classes of entities denoted by the preferred terms both explicitly and unambiguously.

17The skos:definition property is used to define conceptual entries (98% of these) in natural language and in both French and English. Either definitions are original or are taken from the scientific literature, while consensus definitions are preferred if available. For example, a group of experts (Quesque et al. [2024]) selected the definitions for the concepts « mentalizing » and « theory of mind », and their choices have been integrated into the thesaurus. Memory experts are also sometimes requested to revise certain definitions – and more generally, the whole of a given thesaurus entry – and their work is flagged with the inist:reviewedBy property. The skos:scopeNote property can be used to add a note to complete or clarify a definition.

18CogMemo lists over 300 memory phenomena or empirical effects such as the “primacy effect”, « recency effect », « survival effect » and « misinformation effect ». These indicate general and stable features of how memory functions that scientists attempt to explain (Maier et al. [forthcoming]). However, various different factors can sometimes favour, cancel or reverse their appearance (stimulus presentation method, retention interval, experimental design, age of participants, type of memory test, etc.). The inist:moderatorVariable property was created especially so these factors could be documented.

19A selective bibliography is associated with most CogMemo entries (90%) to guide users through scientific literature on memory and help them complete or update their knowledge of concepts. A property for citing a bibliographic reference is not included in the SKOS model so, to make up for this, the thesaurus has incorporated the dct:isReferencedBy property from the Dublin Core7, a standardized vocabulary for describing documents. Details are given of the type of publication study – conceptual analysis, empirical study, historical study, literature review, mega-analysis, meta-analysis, replication, simulation study, software description, test description – and whether access to the document is closed, free or open for all cited publications.

20To take open science and the replicability of research results into account, datasets are regularly added to the CogMemo entries, mainly by drawing on the Open Science Framework8. The le:datasetCitation property from the Linked Earth Ontology is used for this.

21CogMemo entries are also associated with collections which group them by scientific field, study population (human, non-human animal) and according to the theme of ‘Memory and Law’. Finally, image-based elements like graphic representations of research results, timelines, brain structures or summary tables are provided to complete the documentation of certain thesaurus entries.

2.3. Semantic relationships

22The meaning of a concept can also be revealed by its semantic links with other concepts. Hierarchical relations, comprising parent and child concepts, thus make it possible to specify the way in which the entities of a knowledge domain are divided up and thus organized in relation to each other. These relations provide mechanisms for conceptual generalization and browsing in a thesaurus at different levels of specificity. In CogMemo, hierarchical relationships of the genus-species type are in the majority and prioritized. However, the use of partitive and instance relationships is permitted (see Table 3). Hierarchical structuring is implemented using the skos:broader property, to indicate a concept that has a broader meaning than another (such as a hyperonym), and its inverse property skos:narrower, to designate a concept whose meaning is narrower than another (such as a hyponym). The use of polyhierarchy is limited (39 concepts).

Table 3. Examples of hierarchical relationships in CogMemo

Genus-species relationship

Partitive relationship

Instance relationship

cognitive process

medial temporal lobe

patient

memory process

hippocampus

patient H.M.

23The top concepts are placed at the root of the thesaurus and do not themselves have broader concepts, acting, in a way, as gateways into the CogMemo hierarchy. Concerns about conceptual clarification, such as distinguishing memory as a disposition, process, phenomenon or phenomenological property, drove their choice, as did the need to limit and stabilize their number (six in total). These were mostly inspired by the Basic Formal Ontology (Arp et al. [2015]) and the Information Artifact Ontology (Smith & Ceusters, [2015]). The selected top concepts propose a general, more abstract taxonomic framework that is independent of the domain and facilitates the conceptual enrichment of the thesaurus and integration of new related domains and sub-domains, thus diminishing the risk of creating conceptual silos. The top concepts selected are as follows:

  • disposition: “A disposition is the tendency of a capability to be exhibited under certain conditions or in response to a certain stimulus (trigger).”9 This entry covers the main cognitive functions and abilities like learning, attention, emotion, language or memory and also cognitive disorders;

  • information entity: an entity that is about something such as measurements, theories, models, graphs and so forth;

  • material entity: “A material entity is a physical entity that is spatially extended, exists as a whole at any point in time and has mass.”10;

  • phenomenon: “Phenomena are general and stable features of nature, which scientists seek to explain [...]. We tend to identify phenomena as general patterns in data – structures that are observed across datasets – which, in psychology, are often called eects.’” (Maier et al. [forthcoming]);

  • process: “an occurrent entity that exists in time by occurring or happening, has temporal parts, and always depends on some (at least one) material entity.” (Arp et al. [2015: 121]); cognitive, developmental, emotional and neurophysiological processes are hyponyms for the concept “process”; planned processes cover memory study methods and treatments for memory disorders.

  • quality: “A quality is an attribute that is intrinsically associated with its bearer (or its parts), but whose presence / absence and observed / measured value may vary.”11 . Phenomenological characteristics are hyponyms of the top concept “quality”.

24The symmetrical property skos:related is used to specify associative relationships, which are non-hierarchical relationships between concepts. For example, the « episodic memory » concept is associated with the « encoding specificity principle » concept, and therefore, « encoding specificity principle » is associated with « episodic memory » by inference.

25Recent versions of CogMemo have included several newer inverse properties for explicitly mentioning certain types of association. For example, « sentence verification task » is a study method of « semantic memory », therefore, by inference, « semantic memory » has for study method « sentence verification task ». Table 4 lists all these inverse properties.

26Via their URI, CogMemo concepts are aligned with those from other terminology repositories like thesauri, ontologies, online encyclopaedias, by using the skos:exactMatch property which indicates that a thesaurus entry is conceptually identical to an entry in another repository. Alignments therefore constitute an important building block to ensure semantic interoperability between knowledge organization systems. Currently, CogMemo is aligned with eleven repositories – the Cognitive Atlas (cognitive sciences), the SAGE thesaurus (social sciences), MeSH (medicine), SantéPsy (psychiatry and mental health), the UBERON ontology (cross-species anatomy), the Foundational Model of Anatomy ontology (human anatomy) and the Neuro Behavior Ontology (behaviour and behavioural phenotypes in non-human animals and humans). It is also aligned with the participative encyclopaedias Scholarpedia, Wikipedia, wikidata and the FRANCIS philosophy vocabulary.

Table 4. Inverse properties between concepts created in CogMemo

isComponentOf

hasComponent

isDiagnosticToolOf

hasDiagnosticTool

isImpairedIn

isDisorderOf

isMeasureOf

hasMeasure

isModelOf

hasModel

isStudyMethodOf

hasStudyMethod

isTheoryOf

hasTheory

27Before being published online, a series of procedures are run to check that CogMemo conforms to the SKOS model. Among other things, these check there are no redundant or circular hierarchical relationships, that a preferred label exists in each language, that no multiple preferred labels exist for the same language or that the same label has not been duplicated as a preferred, alternative or hidden label.

3. CogMemo: uses

28The following outlines three possible uses for CogMemo: using the thesaurus as an indexing and annotation tool, as an educational repository on memory, and as a source of information to help transfer knowledge to the field.

29Manually or automatically indexing and annotating documents are the classic uses of thesauri. For example, two corpora of bibliographic references on memory, extracted from the Istex archive, have been automatically indexed with CogMemo. One covers memory in psychology and neuroscience12 while the other deals with memory in the humanities and social sciences13 (education, history, literature, philosophy and sociology). The memory concepts in the documents taken from these bibliographic repositories are therefore given a homogenous presentation using a controlled vocabulary, which makes it easier for users to browse through the corpus to retrieve information. Other terminology repositories were used to complete the indexing for fields that CogMemo does not cover well or at all. Obviously, this way of using CogMemo is not limited to scientific articles. Datasets14, data management plans, brain maps, etc., can also benefit from using CogMemo similarly for tagging or indexing concepts.

  • 15 Special thanks to Véronique Reszetko for compiling the citations.

30CogMemo includes definitions, notes, moderator variables and images and indicates semantic relations between concepts which makes it an educational and communication tool on the subject of memory. Informal feedback from users shows that academics and researchers use it as a conceptual and terminological reference for their students. Also, an initial (unpublished) analysis of CogMemo15 citations on the Web suggests that the thesaurus is appreciated for referring readers to definitions of memory concepts, in an academic context or for a more general audience.

31CogMemo also has an interesting potential for transferring knowledge obtained in a laboratory context in the field. Here, the role of memory in a judicial environment can serve as a paradigmatic example. A number of surveys run in different countries have shown that jurors and other persons involved in the judicial world such as experts, magistrates or police officers, often have prior misconceptions about memory and how it functions (e.g. Chun et al. [forthcoming]; Dodier et al. [2019a]; Dodier et al. [2019b]; Magnussen et al. [2010]). In this context, Dodier et al. [2023: 206] particularly recommend that the different parties involved in a judicial investigation should be given better training in the cognitive sciences of memory. To what extent can CogMemo act as an effective memory information resource in the forensic field? Firstly, CogMemo gathers 160 concepts from the thesaurus in a collection that is specially dedicated to the theme of memory and law, which makes it easier for users to access the concepts and terms in the field. Also, CogMemo’s controlled vocabulary provides a common language that should promote communication between the various stakeholders in the sphere of justice. Finally, the documentation of concepts and semantic relationships provides a wide-ranging set of information about memory that can be used in a forensic context.

32Recently, Henry Otgaar and colleagues (Otgaar et al. [2022]; Otgaar et al. [2023]) have suggested that for research results to merit discussion in a courtroom they should fulfil three criteria – replicability, generalizability and practical relevance. Is the CogMemo thesaurus a possible source of information according to these three criteria? Citations of replication studies and datasets help users in terms of replicability of research results. Moderator variables are included in certain entries and can help identify the results’ level of generalizability, while providing more detailed and nuanced information on how memory functions. For example, for the concept « weapon focus effect », one of the moderator variables in CogMemo indicates that the presence of a weapon impairs the memories of the perpetrator and other details of a crime scene more in laboratory or crime simulation studies than in field or archival studies (actual crimes).

33With regard to practical relevance, bibliographic citations of meta-analyses in CogMemo should facilitate for the user the identification of the effect sizes associated with the results of studies on memory. A meta-analysis is a quantitative summary of research findings on the same topic, and an effect size is a measure of the magnitude or strength of a phenomenon. However, Otgaar and colleagues (Otgaar et al. [2022]; Otgaar et al. [2023]) consider the smaller effect size of interest to be a more useful guide for assessing the practical relevance of research findings. For example, the smaller effect size of interest would refer to any factor that causes the gain or loss of at least one piece of information in an eyewitness testimony. Currently there are very few studies that analyze the results of memory research from this standpoint, but in the future this information could be integrated into CogMemo.

4. CogMemo: difficulties and limitations

4.1. Terminological difficulties and limitations

34Managing equivalences between French and English is the main terminological obstacle when editing CogMemo. For example, the French terms MÉMOIRE and SOUVENIR respectively denote the capacity to encode, store and retrieve information and a person’s representation of a past event. However, the term MEMORY generally denotes both these meanings in the English-language literature in cognitive psychology. The term REMEMBRANCE is rarely used in psychological publications. The English term SOUVENIR generally refers to an object which evokes a person, place or journey.

35Currently, CogMemo has not made a definitive decision on how to manage this difficulty. One possible solution would be to use the singular form to denote the different types of memory as dispositions (e.g. MEMORY, AUTOBIOGRAPHICAL MEMORY, EPISODIC MEMORY), with the plural used for terms that denote the different forms of representation of the past and are considered as countable expressions (e.g. MEMORIES, AUTOBIOGRAPHICAL MEMORIES, EPISODIC MEMORIES).

36English is the main language of publication for work on the cognitive psychology of human memory and, consequently, some English terms do not yet exist in French, at least to our knowledge. Also, when translations of English terms are available, the terminology is not always homogeneous and is not stabilized. For example, the French translations of TRANSFER-APPROPRIATE PROCESSING are TRAITEMENT APPROPRIÉ AU TRANSFERT, CONCORDANCE DES TRAITEMENTS and TRANSFERT APPROPRIÉ DE PROCESSUS. Finally, some authors writing in French prefer to use English terms like BINDING or CHUNK in papers.

37In such cases, CogMemo aims to provide French equivalents of English terms through literal translations, with other possible translations indicated as alternative labels. However, a literal translation is not always possible as is the case of the term FORGET-IT-ALL-ALONG EFFECT for which CogMemo has created the French term EFFET D’OUBLI D’UNE RÉCUPÉRATION ANTÉRIEURE.

38In certain disciplines rules exist for writing terms (for example, animal and plant species). It is not CogMemo’s role to reform the vocabulary of the psychology of memory or make definitive pronouncements on the lexical form of its terms, preferring the rather more pragmatic aim of unambiguously denoting concepts in the field. To achieve this goal, the preferred terms are determined by reusing and, where necessary, adapting the vocabulary present in the literature.

39Nevertheless, the community of psychology researchers might need to take on board the fact that much of its vocabulary is borrowed from everyday language. For example, one study (Umanath & Coane [2020]) showed that naïve individuals do not define the terms REMEMBERING and KNOWING in the same way as many experts on memory would when using the same terms in their instructions for experiments (R / K paradigm). Françoise Parot [2022: 11-12], a specialist in the history and epistemology of psychology, has noted that “everyday language is embarrassing for psychology because it seems to expect [psychology] to approve what everyone thinks they know about how humans function.” (my translation). Also, as the psychologist Gregory Kimble [1994: 512] put it, “[o]ne obstacle to psychology’s scientific aspirations is its language, which is the language of the streets.”

4.2. Conceptual difficulties and limitations

40One of the main conceptual obstacles encountered during the enrichment of CogMemo was the lack of consensual definitions. The vague meaning of certain concepts obviously makes it more difficult – if not currently impossible – to write definitions in the form of necessary and jointly sufficient conditions.

41In some cases, definitions depend on the theoretical orientation of their authors. For example, cognitivist orthodoxy posits that cognition consists of using rules to manipulate abstract and arbitrary symbols of the world, or, in other terms, representations. Cognition takes place exclusively in the brain. Conversely, the notion of representation has been left out of radical approaches (Chemero [2011]) to embodied cognition in which the body and its actions on its environment are constituent elements of cognitive functioning. The definitions are not committed to the same entities which makes them difficult to reconcile. Some authors even claim that any attempt to define cognition is, at least temporarily, impossible. For example, the philosopher Colin Allen has suggested that the level of description of the term COGNITION is too general for it to be possible to give a definition (Allen [2017: 4233]). His colleague Marco Facchin considers that the lack of marks to help decide what cognition is and is not makes it impossible to provide a definition in the form of necessary and jointly sufficient conditions (Facchin [2023]). The same observation applies to other concepts. For example, what are the boundaries between memory and learning, memory and imagination, working memory and attention, attention and consciousness or episodic memory and semantic memory? Which marks enable us to distinguish between cognitive memory and immunological memory or even the shape memory of certain materials?

42The recent definition of episodic memory is also interesting as it is representative of difficulties encountered working on other concepts of memory and of cognition more generally (see Renoult & Rugg [2020] for a detailed analysis of the historical development of the concept of episodic memory). In the early 1970s, the Canadian psychologist Endel Tulving defined episodic memory as the memory of unique events that a person is able to locate in time and space (Tulving [1972]; Tulving & Thompson [1973]). More recently, the same author associates it with autonoetic consciousness, a form of consciousness that enables people to attach memories to themselves and then mentally travelling through time, in other terms, to relive past events and imagine the future (Tulving [1985], [2002]). This update of the definition brings up at least two issues. Firstly, it involves the concept of consciousness whose own definition is far from consensual (Brouillet, [2024]). Secondly, the existence of autonoetic consciousness and mental time travel has been the subject of discussion in the literature as regards their suitability for characterizing episodic memory in non-human animals (Latham et al. [2024]; Suddendorf & Busby [2003]; Suddendorf & Corbalis [2010]; Zentall [2006]). Also, even the very fact that animals can remember objects in time and space is not necessarily an analogue of episodic memory in humans (Boyle [2022]; Keven [2022]; Malanowski [2016]). In fact, in studies of non-human animals the term EPISODIC-LIKE MEMORY is often preferred to EPISODIC MEMORY. Autonoetic consciousness has not been consistently defined in Tulving’s work either and even in humans there is debate about its status as a relevant mark of episodic memory (De Brigard [2024]).

43The hierarchical structuring of concepts brings up other conceptual difficulties for editing CogMemo. There are several existing criteria for classifying memory (Werning & Cheng [2017]) which means choices have to be made. For example, in the domain of long-term memory, CogMemo is based on the hierarchical structure of memory systems put forward by Squire & Zola-Morgan [1988] which remains popular as a framework for the taxonomic organization of memory (Madan [2024]). However, this taxonomy is now the subject of debate (Cubelli [2010]; De Brigard [2023]) and alternative or updated taxonomies of memory systems are emerging (e.g. Michaelian [2016]; Murray et al. [2017]; Sherman et al. [2024]; Zárate-Rochín & Marcela [2024]).

44The case of false memories is emblematic of the way taxonomies can come into conflict with the results of scientific research. This concept commonly covers two different types of false memories – those spontaneously generated by an individual’s cognitive system and those produced under the influence of external suggestions and pressures. CogMemo has opted to use this classification. However, several studies have indicated that the classic tasks that measure these two types of memory errors do not correlate to any great extent with each other, if at all (e.g. Bernstein et al. [2018]; Patihis et al. [2018]). These data suggest that different forms of memory illusions have few common properties, if any, which means that grouping them under the broader concept of false memory needs to be reconsidered. The term FALSE MEMORY is not precise enough, and its use would be merely the expression of linguistic convenience (Bernstein et al. [2018]).

45CogMemo does not decide on the ontological status of memory entities. More specifically, integrating a concept into the repository does not depend on whether its reference relates to a natural kind rather than to a conventional or arbitrary category (Bird & Tobin [2022]; Khalidi [2023a]). CogMemo’s objective is to identify, collect and describe concepts found in the scientific literature. However, a conceptual revision within CogMemo in the future will be able to draw on recent discussions on memory or different forms of memory as natural kinds (e.g. Andonovski [2018]; Colaço [2022]; Cheng & Werning [2016]; Gomez-Lavin [2021]; Michaelian [2011], [2015]; Pöyhönen [2016]) and, more generally, on cognitive ontology – the inventory of the existing fundamental entities of cognition and their taxonomic organization (e.g. Dewhurst [2021]; Khalidi [2023b]; Krickel [2024]; McCaffrey &Wright [2022]).

46CogMemo highlights concepts from the cognitive psychology of human memory which means there are currently gaps in its coverage of related fields. For example, the thesaurus does not list all the psychiatric and neurological pathologies that can lead to memory difficulties. Also, the integration of brain structures, genes, neurotransmitters, enzymes and other biological compounds involved in memory functioning remains to be completed while the factors behind individual differences in memory performance are also insufficiently covered.

5. CogMemo: perspectives

47CogMemo is a dynamic repository that expands as it is updated. New entries are added, and the lexicalization and documentation of existing concepts are updated with alternative and hidden labels, definitions, notes, moderator variables, datasets or images. The creation of new collections of concepts like Memory and Education is planned as is the replacement of associative relationships with more explicit properties, for example to specify the relationships between brain structures and memory systems or processes. Two major changes to the thesaurus are currently under discussion, however, with the first involving its semantic model and the second being to increase the participation of memory experts.

48Lexicalizations of concepts are literal data in the SKOS model, which means metadata cannot be used to describe them. To rectify this issue, a transition from SKOS to SKOS-XL16 (Simple Knowledge Organization System eXtension for Labels) is being studied. SKOS-XL is an extension to SKOS that serves to identify, describe and link lexical entities. Labels become resources identified by a URI, which makes it possible to express explicitly, for example, that a label is an acronym of an expanded form. Similarly, other suitable properties (for example, those from the Lexinfo ontology) can be used to describe terms (such as part-of-speech tagging, grammatical gender or the singular / plural form of lexicalizations), the aim being to make it easier for machines to use CogMemo in automatic indexing and annotation tasks or for text mining.

49The advantage of the SKOS model is that it offers a simple system for organizing knowledge. However, properties have been introduced in CogMemo to enrich the documentation of concepts and the semantic relationships between them to make up for the shortcomings of SKOS. This is not a standard practice and it involves technical validation problems because the usual tools for checking conformity with the SKOS model do not take these properties into account. In the longer term, it is necessary to consider an evolution of the thesaurus towards a more complex and expressive model in the form of a computational ontology (Aimé & Arnould [2021]). A computational ontology is a formal (logical) and explicit representation of the concepts in a knowledge domain and their relationships (see Aimé & Arnould [2021] for a detailed presentation of computational ontologies). More particularly, concepts are no longer defined solely in natural language, but in a logical form that can be used by machines for automatic reasoning (logical consistency of modelling, class inference).

50Finding the right solutions to the conceptual difficulties with definitions and taxonomy discussed above – or at least some of them – will condition the evolution of CogMemo in the form of a thesaurus. Such solutions are also required in the event of its migration towards a computational ontology of memory. Inevitably this work will require the increased and enhanced participation of memory experts.

51CogMemo is intended to become a peer-reviewed repository of memory concepts to fully establish its scientific validity. Setting up an editorial and scientific committee should stimulate collaborative work on the thesaurus and encourage the scientific community to use CogMemo. This committee could work with the Memory Research Network (GDR Mémoire)17 made up of over 80 teams and 800 researchers and doctoral students in France working on the subject of memory. It is an interdisciplinary network covering both the humanities and social sciences (education, history, literature, sociology, philosophy, political science) and the biological, psychological and behavioural sciences. And ultimately, the aim is to merge CogMemo into a repository that brings together all the different sciences involved in the study of memory.

Top of page

Bibliography

Adams, Eryn J., Anh T. Ngyuen & Nelson Cowan. 2018. Theories of working memory: Differences in definition, degree of modularity, role of attention, and purpose. Language, Speech & Hearing Services in Schools 49(3). 340-355.

Allen, Colin. 2017. On (not) defining cognition. Synthese 194(11). 4233-4249.

Andonovski, Nikola. 2018. Is episodic memory a natural kind? Essays in Philosophy 19(2). 178-195.

Aimé, Xavier & Frank Arnould. 2021. Modélisation ontologique & psychologies: Une influence réciproque. Paris: Éditions Matériologiques.

Arp, Robert, Barry Smith & Andrew Spear D. 2015. Building ontologies with basic formal ontology. Cambridge, MA: MIT Press.

Bernstein, Daniel, Alan Scoboria, Lecia Desjarlais & Kendall Soucie. 2018. “False memory” is a linguistic convenience. Psychology of Consciousness: Theory, Research, and Practice 5(2). 161-179.

Bird, Alexander & Emma Tobin. 2024. Natural kinds. In Edward N. Zalta & Uri Nodelman (eds.), The Stanford encyclopedia of philosophy. Metaphysics Research Lab, Stanford University.

Boyle, Alexandria. 2022. Episodic memory in animals: Optimism, kind scepticism and pluralism. In André Sant’Anna, Christopher J. McCarroll & Kourken Michaelian (eds.), Current controversies in philosophy of memory, 189-205. New York, NY: Routledge.

Brouillet, Denis. 2024. La fluence motrice comme prédicteur de la conscience phénoménale. In Yann Coello (ed.), Corps, cerveau et processus mentaux : Les fondements sensori-moteurs de la cognition, 75-82. Paris: Dunod.

Chemero, Anthony. 2011. Radical embodied cognitive science. Cambridge, MA: MIT Press.

Cheng, Sen & Markus Werning. 2016. What is episodic memory if it is a natural kind? Synthese 193(5). 1345-1385.

Chung, Kai Li, Kok Wei Tan & I Ling Ding. Forthcoming. Lay person’s and psychology officers’ beliefs about memory, investigative interviewing and deception detection: Data from Malaysia. Psychology, Crime & Law.

Claparède, Édouard. 1910. L’unification et la fixation de la terminologie psychologique. In Édouard Claparède (ed.), VIe Congrès international de psychologie, Genève, 2-7 août 1909, 467-479. Geneva: Librairie Künding.

Colaço, David. 2022. What counts as a memory? Definitions, hypotheses, and “kinding in progress.” Philosophy of Science 89(1). 89-106.

Cowan, Nelson. 2017. The many faces of working memory and short-term storage. Psychonomic Bulletin & Review 24(4). 1158–1170.

Cubelli, Roberto. 2010. A new taxonomy of memory and forgetting. In Sergio Della Sala (ed.), Forgetting, 35–47. Hove: Psychology Press.

Dewhurst, Joe. 2021. Folk psychological and neurocognitive ontologies. In Fabrizio Calzavarin & Marco Viola (eds.), Neural mechanisms (studies in brain and mind) 17, 311-334. Cham: Springer.

De Brigard, Felipe. 2023. Memory and remembering. Cambridge: Cambridge University Press.

De Brigard, Felipe. 2024. Episodic memory without autonoetic consciousness. Philosophical Transactions of the Royal Society B: Biological Sciences. Royal Society 379(1913). 20230410.

Dodier, Olivier, Annika Melinder, Henry Otgaar, Mélanie Payoux & Svein Magnussen. 2019a. Psychologists and psychiatrists in court: What do they know about eyewitness memory? A comparison of experts in inquisitorial and adversarial legal systems. Journal of Police and Criminal Psychology 34(3). 254-262.

Dodier, Olivier, Frédéric Tomas, Melany Payoux & Benjamin Elissalde. 2019b. Professional experience in investigative interviewing does not guarantee strong knowledge about memory. Psychological Research on Urban Society 2(1). 117-125.

Dodier, Olivier, Mickaël Morlet-Rivelli & Fanny Verkampt. 2023. France lags behind in the application of memory science in the courtroom. Journal of Applied Research in Memory and Cognition 12(2). 203-207.

DuCharme, Bob. 2013. Learning SPARQL: Querying and Updating with SPARQL 1.1. Cambridge: O’Reilly Media.

Facchin, Marco. 2023. Why can’t we say what cognition is (at least for the time being). Philosophy and the Mind Sciences 4.

Gomez-Lavin, Javier. 2021. Working memory is not a natural kind and cannot explain central cognition. Review of Philosophy and Psychology 12(2). 199-225.

Hempel, Carl. G. 1966. Philosophy of natural science. Hoboken: Prentice-Hall.

International Organization for Standardization. 2011. ISO 25964-1:2011 Information and documentation – thesauri and interoperability with other vocabularies – Part 1: Thesauri for information retrieval. https://www.niso.org/schemas/iso25964

International Organization for Standardization. 2013. ISO 25964-2:2013 Information and documentation – thesauri and interoperability with other vocabularies – Part 2: Interoperability with other vocabularies. https://www.niso.org/schemas/

Keven, Nazim. 2022. What does it take to remember episodically? In André Sant’Anna, Christopher J. McCarroll  & Kourken Michaelian (eds.), Current controversies in philosophy of memory, 206-224. New York, NY: Routledge.

Khalidi, Muhammad Ali. 2023a. Natural kinds. Cambridge: Cambridge University Press.

Khalidi, Muhammad Ali. 2023b. Cognitive ontology: Taxonomic practices in the mind-brain sciences. Cambridge: Cambridge University Press.

Kimble, Gregory A. 1994. A frame of reference for psychology. American Psychologist 49(6). 510-519.

Krickel, Beate. 2024. The new mechanistic approach and cognitive ontology-or: What role do (neural) mechanisms play in cognitive ontology? Minds and Machines 34(3). 17.

Latham, Andrew J., Kristie Miller & Rasmus Pedersen. 2024. Mental time travel in animals: The ‘when’ of mental time travel. Philosophical Transactions of the Royal Society B: Biological Sciences 379(1913). 20230398.

Laurence, Stephen & Eric Margolis. 1999. Concepts and cognitive science. In Eric Margolis & Stephen Laurence (eds.), Concepts: Core readings, 3-81. Cambridge, MA: MIT Press.

Logie, Robert, Valérie Camos & Nelson Cowan (eds.). 2021. Working memory: The state of the science. New York, NY: Oxford University Press.

L’Homme, Marie-Claude (2020). La terminologie : principes et techniques (2e éd.). Montréal: Presses de l’Université de Montréal.

Machery, Édouard. 2011. Doing without concepts. Oxford: Oxford University Press.

Madan, Christopher R. 2024. Memories that matter: How we remember important things. New York, NY: Routledge.

Magnussen, Svein, Annika Melinder, Elf Stridbeck & Abid Q. Raja. 2010. Beliefs about factors affecting the reliability of eyewitness testimony: A comparison of judges, jurors and the general public. Applied Cognitive Psychology 24(1). 122-133.

Maier, Maximilian, Noah van Dongen & Denis Borsboom. Forthcoming: Comparing theories with the Ising model of explanatory coherence. Psychological Methods.

Malanowski, Sarah. 2016. Is episodic memory uniquely human? Evaluating the episodic-like memory research program. Synthese 193(5). 1433-1455.

McCaffrey, Joseph & Jessey Wright. 2022. Neuroscience and cognitive ontology: A case for pluralism. In Felipe De Brigard & Walter Sinnott-Armstrong (eds.), Neuroscience and philosophy, 427-466. Cambridge, MA: MIT Press.

Michaelian, Kourken. 2011. Is memory a natural kind? Memory Studies 4(2). 170-189.

Michaelian, Kourken. 2015. Opening the doors of memory: Is declarative memory a natural kind? Wiley Interdisciplinary Reviews: Cognitive Science 6(6). 475-482.

Michaelian, Kourken. 2016. Mental time travel: Episodic memory and our knowledge of the personal past. Cambridge, MA: MIT Press.

Murphy, Gregory (2002). The big book of concepts. Cambridge, MA: MIT Press.

Murray, Elisabeth A., Steven P. Wise & Kim S. Graham. 2017. The evolution of memory systems: Ancestors, anatomy, and adaptations. Oxford: Oxford University Press.

Otgaar, Henry, Mark L. Howe. & Olivier Dodier. 2022. What can expert witnesses reliably say about memory in the courtroom? Forensic Science International: Mind and Law 3. 100106.

Otgaar, Henry, Mark L. Howe & Olivier Dodier. 2023. Realistic guidelines on expert witness work concerning memory. Forensic Science International: Mind and Law 4. 100117.

Parot, Françoise. 2022. L’esprit en héritage: D’où vient l’esprit qui hante la psychologie ? Paris: Éditions matériologiques.

Patihis, Lawrence, Steven J. Frenda & Elizabeth F. Loftus. 2018. False memory tasks do not reliably predict other false memories. Psychology of Consciousness: Theory, Research, and Practice 5(2). 140-160.

Pöyhönen, Samuli. 2016. Memory as a cognitive kind: Brains, remembering dyads, and exograms In Catherine Kendig (ed.), Natural kinds and classification in scientific practice, 145-156. New York, NY: Routledge.

Quesque, François, Ian Apperly, Renée Baillargeon, Simon Baron-Cohen, Cristina Becchio, Harold Bekkering, Daniel Bernstein et al. 2024. Defining key concepts for mental state attribution. Communications Psychology 2(1). 1-5.

Renoult, Louis & Michael D. Rugg. 2020. An historical perspective on Endel Tulving’s episodic-semantic distinction. Neuropsychologia 139. 107366.

Sherman, Brynn E., Nicholas B. Turk-Brown & Elizabeth V. Goldfarb. 2024. Multiple memory subsystems: Reconsidering memory in the mind and brain. Perspectives on Psychological Science 19(1). 103-125.

Smith, Barry & Werner Ceusters. 2015. Aboutness: Towards foundations for the Information Artifact Ontology. International Conference on Biological and Biomedical Ontology 1-5. Lisbon, Portugal.

Squire, Larry R. & Stuart Zola-Morgan. 1988. Memory: Brain systems and behavior. Trends in Neurosciences 11(4). 170-175.

Stellato, Armando, Manuel Fiorelli, Andrea Turbati, Tiziano Lorenzetti, Willem van Gemert, Denis Dechandon, Christine Laaboudi-Spoiden, Anikó Gerencsér, Anne Waniart, Eugeniu Costetchi & Johannes Keizer 2020. VocBench 3: A collaborative Semantic Web editor for ontologies, thesauri and lexicons. Semantic Web 11(5). 855-881.

Suddendorf, Thomas & Janie Busby. 2003. Mental time travel in animals? Trends in Cognitive Sciences 7(9). 391-396.

Suddendorf, Thomas & Michael C. Corbalis. 2010. Behavioural evidence for mental time travel in nonhuman animals. Behavioural Brain Research 215(2). 292-298.

Tulving, Endel. 1972. Episodic and semantic memory. In Endel Tulving & Wayne Donaldson (eds.), Organization of memory, 381-402. New York, NY: Academic Press.

Tulving, Endel. 1985. Memory and consciousness. Canadian Psychology / Psychologie canadienne 26(1). 1-12.

Tulving, Endel. 2000. Concepts of memory. In Endel Tulving & Fergus I. M. Craik (eds.), The Oxford handbook of memory, 33-43. Oxford: Oxford University Press.

Tulving, Endel. 2002. Episodic memory: From mind to brain. Annual Review of Psychology 53. 1-25.

Tulving, Endel & Donald M. Thomson. 1973. Encoding specificity and retrieval processes in episodic memory. Psychological Review 80(5). 352-373.

Uddin, Lucina Q., Richard F. Betzel, Jessica R. Cohen, Jessica S. Damoiseaux, Felipe De Brigard, Simon B. Eickoff, Alex Fornito et al. 2023. Controversies and progress on standardization of large-scale brain network nomenclature. Network Neuroscience 7(3), 864-905.

Umanath, Sharda & Jennifer Coane. 2020. Face validity of remembering and knowing: Empirical consensus and disagreement between participants and researchers. Perspectives on Psychological Science 15(6). 1400-1422.

Werning, Markus & Sen Cheng. 2017. Taxonomy and the unity of memory. In Sven Bernecker & Kourken Michaelian (eds.), The Routledge handbook of philosophy of memory, 7-20. New York, NY: Routledge.

Zárate-Rochín, Alba Marcela. 2024. Contemporary neurocognitive models of memory: A descriptive comparative analysis. Neuropsychologia 196, 108846.

Zentall, Thomas R. 2006. Mental time travel in animals: A challenging question. Behavioural Processes 72(2). 173-183.

Top of page

Notes

1 The typographical conventions are as follows (according to L’Homme [2020]): terms are written in capital (e.g. WORKING MEMORY) with concepts in chevron quotes and lower case (“working memory”).

2 https://loterre.istex.fr/P66/en/.

3 https://www.w3.org/TR/skos-reference/.

4 A detailed presentation of the Semantic Web is not within the scope of this article but the technological building blocks essential to understanding the design of CogMemo are discussed.

5 https://dictionary.apa.org/memory.

6 For example, URI for “working memory” in CogMemo: http://data.loterre.fr/ark:/67375/P66-KK6LXTL8-P.

7 https://www.dublincore.org/.

8 https://osf.io/.

9 http://semanticscience.org/resource/SIO_000014.

10 http://semanticscience.org/resource/SIO_000004.

11 http://semanticscience.org/resource/SIO_000005.

12 https://memoire-neurosciencesv2.corpus.istex.fr/.

13 https://memoire-shs.corpus.istex.fr/.

14 See for example, https://doi.org/10.57745/BFEICM.

15 Special thanks to Véronique Reszetko for compiling the citations.

16 https://www.w3.org/TR/skos-reference/#xl.

17 https://gdrmemoire.com.

Top of page

References

Electronic reference

Frank Arnould, “CogMemo: a standardized, structured and formalized terminological repository on human memory”Lexis [Online], Words about #1 | 2025, Online since 12 May 2025, connection on 09 July 2025. URL: http://journals.openedition.org/lexis/8907; DOI: https://doi.org/10.4000/13wn8

Top of page

About the author

Frank Arnould

CNRS – Institute for Scientific and Technical Information
frank.arnould@inist.fr

By this author

Top of page

Copyright

CC-BY-SA-4.0

The text only may be used under licence CC BY-SA 4.0. All other elements (illustrations, imported files) are “All rights reserved”, unless otherwise stated.

Top of page
Search OpenEdition Search

You will be redirected to OpenEdition Search