Beyond the Rival Paradigms Narrative in Adult Learning: A Hybridised and Harmonised Andragogical Model for the Twenty-First Century

Abstract

Demographic ageing and accelerating technological change have made the learning of adults a strategic priority for Europe, yet adult learning theory remains organised around paradigms—pedagogy, andragogy, and heutagogy—that are routinely presented as hierarchically ordered and mutually exclusive. Drawing on a critical review of the literature on andragogy and heutagogy and a cross-case analysis of sixteen training practices collected across the partner countries of the To Switch project, an Erasmus+ strategic partnership, this conceptual article argues that the ideal–typical comparison of these paradigms has reached an impasse—one symptom of which is a proliferation of derivative paradigm labels—and proposes a ‘hybridised and harmonised’ andragogical model built on their convergences. The model is operationalised through learning contracts and a modular sequence inspired by problem solving in technology-rich environments, dynamised by double-loop learning. Implications for the design of adult learning provision and for the competence profile of professionals in adult education and learning are discussed.

1. Introduction

Two long-term transformations frame the present condition of adult education and learning in Europe. The first is demographic: populations and workforces are ageing, working lives are being extended, and the sustainability of welfare and production systems increasingly depends on the opportunity for adults, including those in the later stages of their careers, to keep learning (Kraiger, 2017; Draghi, 2024). The second is technological: automation, digitalisation and, more recently, the diffusion of artificial intelligence (AI) are reshaping the content of work and the skills it requires at a pace that initial education alone cannot match. Where these transformations intersect, lifelong learning ceases to be a slogan and becomes an organisational and societal necessity; adults must repeatedly acquire new knowledge, new competences and, above all, the meta-competence of learning how to learn (Council of the European Union, 2018). To frame the learning of adults in these functional terms alone, however, would be to capture only half of what is at stake. Adult education is also a matter of citizenship; in a knowledge society, continued access to learning is a right, and one of the principal means by which adults remain included in—rather than excluded from—social, cultural, civic and economic life, most of all for those at greatest risk of marginalisation (Delors, 1996). The model proposed in this article is accordingly situated within this twofold horizon, at once functional and emancipatory, rather than within a narrowly functionalist account of adult learning as human capital formation.

The theoretical framework available to those who design adult learning seems not to have kept pace with this scenario. The field remains organised around a small number of named paradigms—classical pedagogy, the andragogy systematised by Knowles (1973), Knowles et al. (2005), and the heutagogy proposed by Hase and Kenyon (2001)—which a substantial body of literature presents in comparative, ideal–typical terms, as if each constituted a unique and exclusive approach destined to supersede its predecessor. As we argue below, this comparative exercise has produced diminishing returns: the debate circles endlessly around the observation that traditional pedagogical formats are neither effective for, nor desired by, adult learners, without converting that observation into cumulative theoretical progress. Meanwhile, in everyday practice, adult education has long combined principles drawn from all three approaches within a single learning process—precisely what the rival paradigms narrative cannot capture (Marcaletti et al., 2021).

This article takes that gap as its starting point. Its general aim is to develop a conceptual model of adult learning—which we call a ‘hybridised and harmonised’ andragogical framework—that moves beyond the rivalry of pedagogy, andragogy and heutagogy. This overarching aim is pursued through three specific objectives: first, to reject the mutual exclusivity of the three paradigms and to systematise instead the convergences among them that the comparative literature tends to obscure; second, to update the andragogical tradition to the conditions of twenty-first century learning, technology-rich environments, problem-centred skill demands and heterogeneous adult publics; and third, to translate the resulting principles into a procedural map centred on the learning contract and on a modular sequence inspired by the framework for problem solving in technology-rich environments (PS-TRE), so that the proposal can inform concrete training design and its subsequent empirical validation. The model was developed within the To Switch project (TOwards Senior Workers’ Innovative Training CHallenges), an Erasmus+ strategic partnership devoted to the training of adult and senior workers (Marcaletti et al., 2021), on the basis of a critical review of the literature on andragogy and heutagogy and a cross-case analysis of sixteen European training practices.

The article proceeds as follows. Section 2 situates the proposal in its demographic and technological context. Section 3 describes the methodology through which the model was developed. Section 4 reviews the founding principles of andragogy and heutagogy, contrasts them with pedagogy, and examines the limits of their ideal–typical comparison; Section 5 summarises the lessons of the practice analysis. Section 6 presents the four assumptions of the model, its principles and its procedural translation. Section 7 discusses implementation criticalities and boundary conditions, and Section 8 concludes.

2. Adult Learning Under Demographic and Technological Pressure

The European demographic transition is no longer a forecast but an operating condition. Old age dependency ratios are rising across the Union, pension reforms have extended working lives, and organisations face the simultaneous ageing of their internal labour markets and a shrinking inflow of young entrants. Under these conditions, the upskilling and reskilling of adult and older workers become a condition of both organisational performance and individual employability (European Commission, 2020), while learning in later life also sustains personal development, active citizenship and social participation beyond employment (Bjursell, 2019). Yet participation in training remains stratified by age, gender and prior qualification: older workers and the low-qualified, who would benefit most from continued learning, are consistently the least likely to receive it (Canduela et al., 2012; Midtsundstad, 2019), and the evidence shows that the effectiveness of training for older learners depends significantly on how it is designed and delivered (Callahan et al., 2003; Kraiger, 2017).

The technological side of the scenario has been codified in successive frameworks of so-called ‘Twenty-First Century Skills’, which converge on the centrality of critical thinking, collaboration, communication, creativity and digital competence (Care et al., 2018; Griffin et al., 2012), and in the European Digital Competence Framework, which articulates twenty-one competences across information and data literacy, communication and collaboration, digital content creation, safety and problem solving (European Commission, 2019). At the assessment level, the OECD Survey of Adult Skills introduced PS-TRE as a distinct domain, conceived as the use of digital technology, communication tools and networks to acquire and evaluate information, communicate and perform practical tasks (OECD, 2012). In its second cycle, conducted in 2023 and released in December 2024, this domain was reconceptualised as ‘adaptive problem solving’—a construct designed explicitly for the digital environments that adults now routinely navigate and for their capacity to react to unforeseen change and emerging information—and the results confirm that proficiency in these information-processing skills remains markedly uneven, with around one in five adults across the OECD below the most basic level of proficiency in any domain (OECD, 2024). More recently still, the diffusion of generative AI since late 2022 has begun to reshape both the skills adults require and the means through which they learn, supplying on-demand tutoring, feedback and personalised learning pathways (Adarkwah, 2025; Asad & Ajaz, 2024) while raising well-documented concerns about over-reliance and cognitive offloading—the tendency to hand over to the tool the remembering, searching and reasoning that learners would otherwise perform themselves—which recent evidence links to a measurable erosion of critical thinking (Gerlich, 2025).

The implication for theory is direct. The question is no longer whether adults must keep learning, but under which principles their learning should be designed so that it produces not only specific skills but capabilities (Stephenson, 1998)—competences usable in unfamiliar circumstances—and the meta-cognitive capacity to keep learning. If anything, the diffusion of generative AI sharpens this question rather than dissolving it; tools that supply readymade answers raise the premium on the capacity to evaluate information, adapt to changing problems and keep learning—the very capabilities against which the available paradigms of adult learning must be assessed.

3. Methodology

This is a conceptual article. Following Jaakkola (2020), it belongs to the family of theory synthesis and theory adaptation papers; it does not report new primary data but integrates existing theoretical, methodological, and empirical materials into a revised conceptual framework, making explicit the procedure through which the framework was derived. The model presented below was developed in two phases, whose outputs were then triangulated.

3.1. Phase 1: Critical Review of the Literature

The first phase of the model design was a critical review of the literature on adult learning paradigms in the sense proposed by Grant and Booth (2009): an engagement with the field that goes beyond description to appraise the quality of existing contributions and synthesise them into a new conceptual model, rather than a systematic or exhaustive mapping of the evidence. Relevant literature was identified through exploratory searches in the Scopus and Web of Science databases, starting from the descriptor ‘andragogy’; because andragogical and heutagogical principles recurred together in the retrieved sources, the reading was extended to ‘heutagogy’ through conceptual snowballing along citation trails and thematic adjacency, rather than through a predefined search protocol. The resulting corpus—a body of around one hundred contributions comprising journal articles, book chapters and books, complemented by reference works on adult education, demographic change and the training of older workers—was read and thematically analysed to surface recurrent principles and, above all, the points of convergence between paradigms that the comparative literature tends to obscure. Selection was guided by conceptual relevance and saturation of the argument rather than comprehensiveness; the review makes no claim to have covered the field exhaustively, and the contributions discussed in the following sections are illustrative of the patterns identified rather than the yield of a reproducible search. This choice was, moreover, consistent with the applied and operational aims of the To Switch project, within which the review was conducted. The project’s purpose was not to produce a systematic stocktaking of the literature for its own sake, but to distill from it the principles and convergences capable of informing the design of training for adult and senior learners. A critical review—oriented towards appraisal and synthesis rather than exhaustive retrieval—was therefore the approach best suited to that practical end, since it allowed the analysis to move directly from the recurrent patterns in the literature to the operational model developed in the following sections.

To make the procedure transparent while keeping the review exploratory rather than systematic, we add the following details. The review was conducted during the first phase of the To Switch project, in 2020–2021, as a scoping exercise and not as a protocol-driven search; it began from the single descriptor ‘andragogy’ and, once the recurrent association between andragogical and heutagogical principles became apparent, was repeated with the descriptor ‘heutagogy’, and was complemented by reference works on adult education, demographic change and the training of older workers and by snowballing along citation trails. No Boolean protocol, date limit or predefined inclusion grid was applied; contributions were retained on grounds of conceptual relevance to the principles, comparison or design of adult learning, and the corpus was delimited by argumentative saturation rather than by exhaustiveness. Recurrent principles and, above all, points of convergence were identified through iterative thematic reading by the authors, and divergent interpretations were reconciled by discussion until consensus. The review therefore makes no claim to systematicity or reproducibility, and the contributions cited are illustrative of the identified patterns.

3.2. Phase 2: Cross-Case Analysis of Good Practices

The second phase of the model design analysed sixteen training practices addressed to adult and senior learners and their trainers, collected by the project partners in their respective countries. A good practice was defined as a practice that, against explicit criteria, has proved to work well in real settings and to produce results superior to those achieved by other means, so that it can serve as a benchmark. Selection followed seven criteria: the development of new approaches to adult and senior upskilling and reskilling; the adequacy of design and implementation (internal and external consistency of objectives, activities, resources and evaluation); the effectiveness of results and their recognition by the peer community; innovativeness relative to context; sustainability in the medium to long run; reproducibility and transferability; and the potential for mainstreaming into adult training and lifelong learning systems. Each practice was described through a common form covering both descriptive elements and the evidence for each criterion. To enable comparison, the practices were classified into four groups according to the target of the training: practices addressed to trainers, teachers, tutors and human resource managers (nine); practices supporting the upskilling, reskilling and motivation of employed older workers (four); practices addressed to the employability of seniors outside the labour market (three); and practices targeting senior citizens through voluntary and community-based peer learning (two). A transversal cross-reading of the forms identified recurrent methods, contents and lessons learned, as well as systematic gaps.

It should be stressed that these sixteen practices were purposively identified and documented by the To Switch partners in their respective countries against common reporting criteria; they constitute an illustrative, non-representative collection rather than a sample. To limit confirmation bias, the cross-case reading deliberately recorded gaps and negative lessons alongside convergences. The four figures above (nine, four, three, and two) refer to category placements and sum to eighteen placements across the sixteen unique practices, because two practices (NO01 and RO02) each serve two target groups; the practice analysis therefore provides illustrative support for the model rather than empirical validation.

3.3. Integration

The model presented in the following sections results from the integration of the two phases; the assumptions were derived from the convergences identified in the literature, tested for plausibility against the evidence of the practices, and then translated into procedural terms. The limitations of this procedure—above all the fragmentation and uneven quality of the available literature, and the documentary nature of the practice evidence—are discussed in Section 7.

4. Andragogy, Heutagogy and Pedagogy: Principles and the Limits of Their Comparison

4.1. Andragogy

Although the term predates him by more than a century (Loeng, 2017), and although classical pedagogy was for centuries the only recognised educational paradigm before andragogy was articulated as its adult counterpart (Ozuah, 2016), andragogy owes its systematisation to Knowles, for whom it denoted the art and science of helping adults learn, in contrast with a pedagogy oriented to children (Knowles, 1973, 1978). In its mature formulation, andragogy is explicitly a process model rather than a content model: it specifies a set of principles for designing the learning process of adults, whatever the content to be learned (Knowles et al., 2005). Six core principles are identified: the adult’s need to know the why, what and how of learning before undertaking it; a self-concept characterised by autonomy and self-directedness; the role of prior experience as both a resource for and an occasionally obstructive mental model in learning; a readiness to learn linked to life tasks and changing social roles; a problem-centred orientation towards the immediate application of knowledge; and a motivation to learn that is primarily intrinsic, tied to personal value and payoff rather than external reward (Knowles et al., 2005; Merriam, 2001; Marcaletti et al., 2018). The principles have proved durable and broadly transferable, informing training design well beyond their origins and feeding explicit programmes for building sustainable lifelong learning societies (Charungkaittikul & Henschke, 2017; Henschke, 2011).

Andragogy has also accumulated a familiar catalogue of criticisms: that it homogenises adult learners and ignores variation among them (Brookfield, 1986); that it privileges process over content and intrinsic over extrinsic motivation (Davenport & Davenport, 1985); that it is an ideology or a set of practical principles rather than an empirically grounded theory (Hartree, 1984); and that the assumptions of andragogy sit awkwardly with the prior conditioning of many adult learners, whose earlier schooling has accustomed them to depend on the teacher rather than to direct their own learning (Arghode et al., 2017; Loeng, 2018). The status of andragogy as a testable theory has been a particular target. It has been characterised as a set of plausible but loosely connected propositions that has yet to be placed on a firm empirical footing (Rachal, 2002), and efforts to turn it into a validated instrument for measuring andragogical practice have met only partial success (Holton et al., 2009). Critical readings have gone further, questioning whether the self-directedness that adult education has come to take for granted, and to build into its programmes, actually delivers the autonomy it promises (Kruszelnicki, 2020). These criticisms matter for what follows. They weaken the case for treating andragogy as a closed, self-sufficient paradigm—but not the practical value of its principles, which is what our proposal retains.

4.2. Heutagogy

Heutagogy, proposed by Hase and Kenyon (2001) and developed by Blaschke and Hase (2016) among others, presents itself as the study of self-determined learning; described by its originators as ‘a child of complexity theory’ (Hase & Kenyon, 2007), meaning that it treats learning as an unpredictable, self-organising process rather than a linear progression, it was designed for capable people operating in turbulent, fast-changing environments. It radicalises the andragogical principle of self-direction: the learner becomes the primary agent who decides what will be learned, how, and how it will be assessed. Five principles define the approach: learner-centred and learner-determined learning grounded in human agency; capability, understood as the ability to use competences in unfamiliar as well as familiar circumstances, combining self-efficacy, communication, creativity, collaboration and positive values; self-reflection and metacognition, whereby learners reflect not only on what they learn but on how they learn it (Flavell, 1979); double-loop learning, in which learners question the values and assumptions underlying their problem solving (Argyris & Schön, 1978; Blaschke, 2012); and non-linear learning paths defined by the learner rather than the teacher (Blaschke & Hase, 2016). The educator is correspondingly redefined as a facilitator and ‘knowledge broker’ rather than a transmitter (Ashton & Newman, 2006), and the radical autonomy the approach grants the learner has even been read through the lens of free will (Shpeizer & Glassner, 2020). A distinctive feature of heutagogical design is the negotiated, individualised learning contract, through which learner and educator jointly fix objectives, methods and assessment. The approach has been applied across a wide range of fields—including nursing and health education; the development of management and adaptive capability; mentoring and teacher education; and community- and society-based learning (Bhoyrub et al., 2010; Halsall et al., 2016; Levy-Feldman, 2018; Stoten, 2021), among others—and reviews document both its growing uptake in technology-mediated settings and the persistent thinness of its empirical base (Agonács & Matos, 2019; Moore, 2020). More recent work has tied heutagogy explicitly to digital technology and lifelong learning (Blaschke, 2021; Hase & Blaschke, 2021) and has begun to reconceptualise self-determined learning for an environment saturated by generative AI, where the decisive variable becomes how the learner relates to AI as a source of knowledge rather than the technology itself (Alm, 2024; Levenberg & Bitar, 2026).

4.3. Pedagogy

Pedagogy, by contrast, is the older and still dominant paradigm against which both andragogy and heutagogy were originally defined. Etymologically defined as the art of leading the child, it long stood as the only recognised educational model, and the very coinage of the term ‘andragogy’ was a deliberate attempt to mark its limits for adult learners (Knowles, 1973; Ozuah, 2016). In its classical form, pedagogy is at once a content model and a dependency model: the teacher holds authority over what, how, and when something is learned and how it is assessed; learners are positioned as dependent recipients; curricula are standardised; and knowledge is transmitted unidirectionally from those who hold it to those who lack it (Knowles et al., 2005). Motivation is assumed to be largely extrinsic, driven by external reward; learning operates at the cognitive rather than the meta-cognitive or epistemic level; and the values and assumptions underpinning what is taught are not themselves placed in question, so that change in the double-loop sense is absent (Argyris & Schön, 1978). It is precisely this configuration that the adult-learning paradigms set out to invert, and making it explicit clarifies the terms of the comparison that follows.

4.4. The Ideal–Typical Comparison and Its Impasse

A large share of the literature proceeds by comparing the three approaches in ideal–typical terms extracting the characteristics of each in their pure state and arraying them along a progression from dependency through self-direction to self-determination in education (Bansal et al., 2020; Blaschke, 2012; Jones et al., 2019), a progression often presented as a natural developmental sequence from andragogy to heutagogy (Chacko, 2018). Table 1 summarises this comparative ideal–typical description across eleven features extracted from the literature, from target learners and learning objectives to motivational factors and the place of creativity and collaboration.

Table 1. Comparative description of traditional teacher-centred pedagogy, andragogy and heutagogy in the ideal–typical literature.

Feature	Pedagogy	Andragogy	Heutagogy
Target learners	School-age pupils or inexperienced students	Adults with or without experience	Adults with some experience
Objective of learning	Acquiring knowledge to reach the next stage (acquisition process)	Developing skills to solve specific problems (personal development linked to vital needs)	Developing capabilities based on the need and potential to learn (transformative process)
Organisation of learning	Standardised curricula; teacher-centred	Needs-based; learning how to learn, with or without the teacher; problem-centred contextual model	Not standardised; learner-centred
Role of learners and teachers	Learners dependent; teachers decide what, how and when (dependency model)	Learners autonomous; teachers guide and facilitate self-direction (self-direction model)	Learners independent; educators stimulate curiosity and offer opportunities (self-determination model)
Learning process	Unidirectional, teacher to learner	Bidirectional, teacher and learner	Multidirectional
Learning resources	Limited: designed by teachers	Controlled: decided collaboratively	Unlimited: mainly decided by the learner
Level of cognition	Cognitive	Meta-cognitive	Epistemic (evidence-based)
Assumptions on change	Absent (single-loop)	Absent (single-loop)	Present (double-loop)
Motivational factors	External, driven by rewards	Internal, driven by needs (intrinsic value, payoff)	Internal, driven by reflexivity
Allows creativity	No	No	Yes
Requires collaboration	No	Not essential	Yes

Source: authors’ elaboration based on (Arghode et al., 2017; Bansal et al., 2020; Blaschke, 2012; Knowles et al., 2005).

Table 1 should be read as an ideal–typical reconstruction, in the Weberian sense (Weber, 1949), of how the comparative literature portrays the three approaches in their pure state, and not as a description of pedagogy, andragogy or heutagogy as they are practised today. In particular, the column for pedagogy refers to traditional teacher-centred pedagogy; contemporary pedagogy—including inquiry-based, collaborative, project-based and learner-centred instruction—already shares many of the features the ideal type assigns to andragogy or heutagogy. That these features cut across the supposed categories is not incidental to our argument but part of it: it is a first indication that the ideal–typical distinctions are overdrawn, as the remainder of this section contends.

Read across its rows, Table 1 lays out a consistent gradient rather than a set of unrelated contrasts. As one moves from pedagogy through andragogy to heutagogy, the learner shifts from dependent to autonomous to independent; the object of learning moves from acquiring knowledge to developing skills to developing capabilities; and the educator recedes from controller to facilitator to a resource the learner may or may not consult. The same progression runs through the deeper features: cognition deepens from the cognitive to the meta-cognitive and finally the epistemic level; the assumption on change passes from absent (single-loop) to present (double-loop); motivation is recast from extrinsic reward to intrinsic need and then to reflexivity; and creativity and collaboration, treated as inessential at the pedagogical pole, become central at the heutagogical one. It is the very neatness of this gradient that makes the ideal—typical reading attractive—and, as the rest of this section argues, misleading.

The comparison is pedagogically useful, but as a research programme it has reached an impasse, for three reasons (Marcaletti et al., 2021). First, it overstates divergence by default: ideal types push each approach to its extreme, so that differences are amplified, and convergences disappear from view (Weber, 1949). Second, it has not produced cumulative knowledge: the conclusion that traditional pedagogical formats are neither effective for nor desired by adult learners is rediscovered repeatedly without advancing the theory. Third, a substantial portion of this literature—including comparative empirical studies published in indexed journals—exhibits conceptual inaccuracies and methodological weaknesses serious enough to undermine its evidential value; the field’s quality control and peer review mechanisms have not reliably filtered them. We level this criticism at the general tendency rather than at specific publications by name: the weakness we are describing is a systemic feature of this strand of the literature, not the failing of a few identifiable authors.

A further symptom of the impasse is terminological rather than empirical. Rather than integrating the established approaches, the field has tended to respond to each new context by coining yet another paradigm: academagogy for organisational and corporate learning (Murthy & Pattanayak, 2019); ubuntugogy and ‘egonagy’ for African and communal educational traditions claimed to transcend pedagogy, andragogy and heutagogy alike (Bangura, 2005); cybergogy and e-heutagogy for digital and online settings (Canţer, 2012; Sumarsono, 2020); and digital or technology-infused variants of andragogy, sometimes formalised as ‘technology–andragogy content knowledge’ models (Arifin et al., 2020; Blackley & Sheffield, 2015). The accumulation of these neologisms has itself become an object of comment (Peterson & Ray, 2013), and surveys suggest that practitioners are frequently unclear about what distinguishes the resulting labels (Brieger et al., 2020; Tümen Akyıldız, 2019). This proliferation is precisely the conceptual inflation our proposal seeks to halt: multiplying names for adjacent ideas obscures their common ground and substitutes labelling for cumulative theory.

Against this background, two strands of evidence are more productive. The first is the meta-analytic finding that training method matters for older learners’ performance; design choices, not age as such, drive much of the variance in outcomes (Callahan et al., 2003). The second is the stream of work that treats critical and computational thinking applied to problem solving in technology-rich environments as a transversal, teachable process: at each phase of a problem-solving cycle, reflection on impacts allows the learner to return to the initial problem and redefine it, which is precisely where double-loop and multiple-loop learning operate (Íñiguez-Berrozpe et al., 2020; Íñiguez-Berrozpe & Boeren, 2020). Vanek (2017) has translated the PIAAC PS-TRE framework into phases and activities directly usable in adult education. These strands suggest that what adult learning theory needs is not a further paradigm to set against the existing ones, but a principled way of combining them around evidence-based processes.

5. Lessons from Practice

A cross-case analysis of the sixteen practices corroborates this suggestion (see Table 2 and Appendix A). The sixteen practices were grouped, according to the population they addressed, into four categories: practices on training methodologies aimed at the trainers, teachers, tutors and HR managers of adult and senior learners; practices supporting the upskilling, reskilling and motivation of employed senior workers; practices fostering the employability of seniors outside the labour market; and practices targeting senior citizens through voluntary involvement, community engagement and peer learning. Across these four target groups, the practices deploy a wide repertoire of models, methodologies and tools—frequently invoking andragogical principles and combining them in blended and collaborative formats with needs assessment at entry and, in the more structured cases, validation or certification of acquired skills at exit. Three transversal elements recur: collaborative and peer-based approaches, including intergenerational and community-based learning; blended methodologies that integrate face-to-face and digital delivery; and attention to the assessment of needs and of acquired competences as the hinge between training and its recognition. In none of the documented practices does strict adherence to a single named paradigm appear as a design criterion: combination is the norm, not the exception.

Table 2. The sixteen good practices grouped by target population.

Target Group (Practices)	Predominant Methods and Approaches	Principal Contribution to the Model
Trainers, teachers, tutors and HR managers (CH01, DEN02, ES01, IRE02, ITA01, ITA03, RO01, RO02, NO01)	Train-the-trainer and project work designs; experiential and blended learning; needs and competence assessment; learning/training pacts; intergenerational and mentoring approaches	Update of adult training paradigms and, above all, definition of the adult-trainer competence profile; materials and structures for the digital platform
Employed senior workers (DEN01, IRE01, ITA02, RO02)	Andragogic personal development training combined with job-specific, digital and soft skills; dual (classroom + workplace) and blended models; counselling and self-assessment; life-course perspective	Enhancement of self-direction and motivation; validation of competences; preventive life-cycle approach
Senior unemployed and inactive (FR01, FR02, CH02)	Small-group, highly adaptive designs combining motivation and self-confidence building with job/sector-specific and entrepreneurial skills; co-development and mentoring	Motivation and voluntary adherence; work on self-image and stereotypes; flexible, learner-centred support
Senior citizens (ES02, NO01)	Dialogic, collaborative and peer-based learning; community involvement and volunteering; senior-friendly digital upskilling	Democratic participation and inclusion; intergenerational and community learning; digital inclusion of seniors

Note. Sixteen practices collected across eight European countries within the To Switch project; see Appendix A for the full description of each. NO01 and RO02 were each delivered to two target groups and therefore appear twice.

Equally instructive are the systematic gaps. The practices rarely adopt a life-course approach in which lifelong learning is articulated through age-specific training strategies; they almost never define the competences and professional profile of the trainers involved, nor methods for validating and certifying those trainers’ skills; and their sustainability is fragile, with many initiatives dependent, for instance, on European funding and likely to lapse when it ends—a weakness particularly visible in countries lacking a structured mainstream training system for adult and senior populations. A new conceptual proposal, to be useful, must therefore not only state learning principles but support their procedural implementation and the definition of the trainer profile that implementation requires. Nor can its usefulness be assessed at the level of individual courses alone: if it is to counter the fragility and funding-dependence that the practice analysis brought to light, such a proposal must also be capable of being embedded in the policies and mainstream systems of adult education and learning—so that its principles inform the design, provision, quality assurance and recognition of adult training as a matter of ordinary provision rather than of time-limited, project-based initiatives.

6. A Hybridised and Harmonised Andragogical Model

A paradigm is at once a pattern or model that organises how something is understood and done, and the broader framework of shared assumptions, concepts and values from which the theories, methods and practices of a field derive their coherence (Kuhn, 1962). The proposal that follows is intended in both senses: a model that develops the andragogical approach and adapts it to the needs and characteristics of twenty-first-century adult learners, and a vision of adult learning that applies the available scientific evidence with greater rigour than the rival paradigms narrative allows. As a consequence, the proposed framework rests on four assumptions.

6.1. Four Assumptions

The first assumption is convergence rather than divergence. Ideal–typical classifications take the characteristics of each approach to their extremes and thereby amplify differences; we hold, on the basis of the practices analysed, that it is more revealing, and analytically more productive, to insist on the common principles and convergence factors among pedagogy, andragogy and heutagogy. Earlier responses to the same intuition have tended either to coin a further, supposedly transcendent paradigm or to call broadly for ‘new models’ of lifelong learning (Bangura, 2005; Kanwar et al., 2019); our aim is the opposite—to integrate the established approaches rather than to add to their number. No empirical evidence establishes these approaches as univocal or mutually exclusive.

The second assumption is that andragogy already incorporates self-direction, the very principle that heutagogy elaborates into self-determination (Blaschke, 2012). The semantic distinction between self-directed and self-determined learning is subtle and, while conceptually meaningful, of limited consequence for the design of learning—indeed, the founding heutagogical literature itself sometimes uses the terms interchangeably. Both reduce to freedom, autonomy and independence of judgement on the part of the learner, and we prefer these terms for their superior semantic quality. Far from differentiating the approaches, they reveal their continuity.

The third assumption locates the unifying device of adult learning design in the learning contract. The key starting point of a heutagogical process is the joint definition, on the basis of independent judgement of the learner and educator, of an individualised learning contract that fixes the learner’s preferences regarding what, how, when and where to learn (Knowles, 1986). But nothing in the logic of self-determination excludes that the design fixed in the contract incorporates, for given contents, principles of traditional pedagogy, of andragogy, or of heutagogy: a learner exercising full autonomy may rationally prefer direct instruction for one component, tutored exercise for another, and self-determined project work for a third. What self-determination governs, then, is the learner’s freedom to choose which learning principles to adopt for each part of their learning; it does not require that this choice fall on a single paradigm, nor that the same principles apply throughout.

The fourth assumption is that hybridisation reflects what actually happens in practice. In real learning processes—in adult education, but also in higher education, vocational training or organisational upskilling, among others—what is achieved is a mix of principles, whose effectiveness depends on the preparation of the educator and on the motivation and self-directedness of the learner alike. Consider an adult who needs to learn spreadsheet-based software for statistical analysis. The learning contract may legitimately combine, without loss of coherence: a traditional, unidirectional module in which software functions and commands are explained; a module of andragogical inspiration in which the learner, tutored by the educator, exercises the software against personal needs; and a project work module of heutagogical inspiration in which the learner conducts an original analysis with the software, supervised but self-determined, in response to a personal need or desire. The hybridised and harmonised framework is, in short, a convergent model: it recognises the autonomy and independence of judgement of the adult learner in formalising learning contracts within which the hybridisation of pedagogical, andragogical and heutagogical principles is rewarded, in relation to the learner’s previous experience, preferences, needs and personal payoffs.

6.2. Principles of the Model

Table 3 restates the eleven comparative features of Table 1 as principles of the hybridised model, identifying for each the point of convergence that benefits adult learning practice.

Table 3. Principles of the ‘hybridised and harmonised’ andragogical model.

Feature	Hybridised and Harmonised Principle
Target learners	Adult and senior learners, whatever their prior life, work and learning experience
Objective of learning	Acquiring new knowledge and developing new competences and capabilities, based on the need to solve new problems, including the meta-cognitive ‘learning how to learn’ skill
Organisation of learning	Modular, combining standardised curricula, self-directed and self-determined learning
Role of learners and teachers	Learners and teachers are autonomously co-dependent, in learning and assessment, within the limits of the learning contract they have set (reciprocal bipolar model)
Learning process	Multidirectional and open to all actors: knowledge flows not only from teacher to learner but among learners, peers, stakeholders and the wider social environment and community
Learning resources	Multilevel, evidence-based, drawn from trustworthy sources
Level of cognition	Cognitive, meta-cognitive and epistemic; critical and computational thinking addressed to problem solving and directed to change
Assumptions on change	Double-loop and multiple-loop learning as default
Motivational factors	Holistic, reflexivity-based, melding extrinsic and intrinsic value and personal payoff
Allows creativity	Yes, at every learning stage
Requires collaboration	Yes, at every learning stage

Source: authors’ elaboration.

Some of these principles deserve brief elaboration. The objective of learning is comprehensive: the acquisition of knowledge that supports the development of competences understood as capabilities—usable in unfamiliar circumstances and oriented to problem solving in complex, technology-rich situations—with the meta-cognitive skill of learning how to learn as its fundamental component. The relationship between learner and educator is conceived as reciprocal and bipolar: the two are autonomously co-dependent within the limits of the contract they have set—the educator recognising and promoting the learner’s autonomy, the learner recognising the value of the educator’s experience, guidance and supervision. The underlying cognition is multiple—cognitive, meta-cognitive and epistemic. In adults, age-related declines in fluid abilities are offset by the stability of crystallised abilities (Marcaletti et al., 2014; Warr, 1994); bringing adult learners to learn how to learn therefore requires stimulating the epistemic level—critical and computational thinking applied to problem solving and fed by reliable evidence—since only at that level can prior values and knowledge be altered and personal and social change be triggered, in the sense long associated with critical and transformative adult learning (Şahin & Doğantay, 2018). Finally, the default assumption about change is double-loop learning—questioning the assumptions behind a problem, not just correcting the response to it—which, when learning repeatedly reaches the meta-cognitive and epistemic levels over a lifelong trajectory, extends into multiple-loop learning. Motivation, correspondingly, is treated holistically, as a reflexive balance of intrinsic and extrinsic value and personal payoff, while creativity and interaction among learners are promoted at every stage rather than reserved for one paradigm.

6.3. Procedural Translation

For principles to carry weight in adult education, they must be translatable into practice. The model is therefore operationalised as a small set of procedural rules that convert its assumptions into design decisions. First, the learning contract is negotiated collaboratively, taking into account not only the learner’s preferences but also the expectations of the other stakeholders in the learning process. Second, learning activities are structured around the PS-TRE process, adopted here as an organising framework because it renders problem solving in technology-rich environments as a teachable sequence and because the PIAAC literature has underlined its relevance to workplace learning (Olsen & Tikkanen, 2018). Third, that sequence is articulated into phases and subphases calibrated to the learners’ prior experience and motivation. Finally, each subphase is matched to the learning principle best suited to it—dependency, self-direction, or self-determination—so that pedagogical, andragogical, and heutagogical principles are combined within a single itinerary rather than applied uniformly across it. Table 4 presents the resulting procedural map: a modular sequence that opens with pre-activity evaluation, proceeds through the adapted PS-TRE phases, and closes with post-activity evaluation of acquired skills and of the social impact of the problem solving achieved; for each step, the procedural map indicates consolidated methodologies and digital resources, many of them drawn from blended and self-directed designs that combine face-to-face and online provision and place the management of the learning path in the learner’s hands (Giuseffi, 2019; McKenna et al., 2020; Youde, 2018). The sequence is dynamised by the double-loop principle: at each phase, reflection on outcomes and impacts can send the process back to the redefinition of the initial problem, objectives and resources, which is what converts a training course into a learning-to-learn experience.

Table 4. Procedural map of the model: PS-TRE-inspired sequence, activities and supporting resources.

Phase	Activity	Examples of Methodologies and Digital Resources
0. Pre-activity evaluation	Assessment and self-assessment prior to training	Tools for assessing prior knowledge and skills (e.g., OECD Education and Skills Online); assessment of previous experience; ex-ante evaluation of method preferences
1. Make a plan	Set goals	Repositories and databases giving access to reliable, evidence-based sources; guidelines and principles for adult learning; methods and contents drawn from good practices
	Identify constraints	Guidelines or digital applications supporting SWOT analysis
2. Subset cases/break down and organise tasks	Plan and organise	Guidelines or digital applications supporting project work and Gantt planning
3. Evaluate information	Acquire and evaluate information	Monitoring and self-assessment guidelines, applications and sheets; tutorials on videoconferencing platforms
4. Monitor progress	Move towards goals	Monitoring and self-assessment guidelines, applications and sheets
5. Use applications	Apply information	Self-assessment guidelines, applications and sheets; presentation tools
6. Post-activity evaluation	Assessment and self-assessment at the end of training	Tools for assessing acquired knowledge and skills (including soft skills) and the social impact of the problem solving achieved

Source: authors’ elaboration, adapting the PS-TRE phases (OECD, 2012; Vanek, 2017) to the hybridised model; the sequence is dynamised by double-loop learning, allowing return from any phase to the redefinition of the initial problem.

The procedural map in Table 4 follows the cognitive dimensions of problem solving in technology-rich environments set out in the PIAAC framework (OECD, 2012; Vanek, 2017), and each phase mobilises specific principles of the three approaches. The pre- and post-activity assessment phases enact the andragogical concern with diagnosis and with evidence-based, learner-centred evaluation; goal-setting and the identification of constraints enact the heutagogical principle of learner-defined objectives, fixed through the andragogical learning contract; planning, information handling and the monitoring of progress enact self-directed and self-determined learning together with the meta-cognitive, double-loop reflection characteristic of heutagogy and, more broadly, of critical and transformative adult learning (Mezirow, 1991, 2000); and the application phase enacts the andragogical, problem-centred orientation to immediately usable knowledge. It is worth noting that, although the ideal–typical comparison in Table 1 associates single-loop learning with andragogy and double-loop learning with heutagogy, our framework deliberately lifts double-loop learning out of that exclusive association and adopts it as a cross-cutting principle that dynamises the whole design—one of the convergences the model harmonises rather than an inconsistency.

7. Discussion

Before turning to the model’s limitations, a word on its scope. Because it is specified at the level of learning principles and as a content-agnostic procedural map, rather than in terms of any particular subject matter, the model is transversal by conception and, in this sense, applicable across contents, publics and settings, though its usefulness may vary across cultural settings, as noted below. Within the To Switch project, it was used to design training in the digital literacy of older people, in adult language learning, and in provision for jobseekers; beyond the project, it was also used to train trainers and line managers in companies. Notwithstanding this transversality of application, three orders of criticality accompany the proposal. The first concerns implementation. The learning contract, the model’s unifying device, is not always feasible in its first best, fully individualised form: in contexts that synchronously involve large groups, personalised objectives and implementation cannot always be defined for each learner. Workable second-best solutions exist—defining areas, methods and objectives common to the group, accompanied by personalised tutoring and feedback, or using project work individually or in small groups under the educator’s supervision—but they must be designed in, not improvised. Similarly, the articulation of modules oriented to distinct principles presupposes an educator able to guide learners towards the most appropriate principle for each module, and learners possessing a reflexive competence about learning methods that cannot be taken for granted in all adults. The double-loop principle mitigates this criticality—it allows gaps to be identified and the process to return to the definition of objectives and resources—but it demands experience from the educator and reflexivity from the learner, and presupposes the modular organisation of the activity. Assessment, finally, must be agreed with learners and communicated with clarity at every stage, since pre-, ongoing, and post-activity evaluation also requires a level of awareness that has to be built rather than assumed.

The second criticality concerns the educator. The model demands a figure who is simultaneously an expert in a field of knowledge, an expert in learning processes able to deploy pedagogical, andragogical and heutagogical principles discriminately, a practitioner of critical and computational thinking applied to PS-TRE, and a designer of evidence-based assessment—closer to a learning partner, counsellor and knowledge broker than to a transmitter (Wahlgren, 2016). The full competence profile of this figure, developed in the same project as a set of ten certifiable core competences, exceeds the scope of this article and is reported in the project’s first intellectual output (Marcaletti et al., 2021).

The third criticality concerns the learner and the model’s boundary conditions. Adult publics are heterogeneous in their motivation to learn, and previous research documents adults who explicitly prefer traditional, lecture-based formats (Íñiguez-Berrozpe & Marcaletti, 2016). The model accommodates this heterogeneity by design: the contract may legitimately fix pedagogical principles wherever the learner prefers them. But adults differ along many dimensions besides motivation, and the same adaptive logic must extend to all of them—to the learner’s employment situation (in work, unemployed or inactive), prior experience, learning needs, age and gender—each of which may call for a different configuration of the learning path. Above all, the model is at this stage a conceptual proposal grounded in a critical review and documentary practice analysis: it awaits empirical validation through the piloting of training activities—that is still ongoing—designed according to its procedural map, including the digital platform developed in the subsequent phases of the To Switch project (Marcaletti et al., 2021). Its claims should be read and tested as design hypotheses.

Taken together, these considerations allow the article’s three specific objectives to be revisited. The first—to reject the mutual exclusivity of pedagogy, andragogy and heutagogy and to systematise their convergences—has been met by re-reading the ideal–typical gradient of Table 1 as a set of complementary principles rather than competing paradigms, and by showing, through the cross-case analysis of practice, that their combination is already the norm in adult education rather than the exception (Marcaletti et al., 2021). The second—to update the andragogical tradition to the conditions of twenty-first-century learning—has been addressed by anchoring the model in technology-rich, problem-centred and lifelong demands: the learning contract absorbs the self-determination that heutagogy foregrounds, while the PS-TRE-inspired sequence and its double-loop dynamic connect andragogical principles to the information processing and adaptive problem-solving skills that current frameworks and assessments place at the centre (OECD, 2024). The third—to translate these principles into a procedural map usable in design and open to validation—has been pursued through the modular sequence of Table 4, framed by pre- and post-activity evaluation and specified at a level of detail that makes the proposal at once implementable and testable. The model thus answers its stated aims, while the criticalities discussed above make explicit the conditions under which those aims can be realised in practice.

It is worth being precise about what is, and what is not, novel here. That autonomy is situational, that learners need different degrees of guidance, and that methods can be combined within a single programme are well established, whether in the literature on self-directed learning (Candy, 1991) and scaffolding, on blended and experiential learning, or—most directly—in the Pedagogy–Andragogy–Heutagogy (PAH) continuum (Blaschke, 2019) and in recent hybrid proposals that combine andragogical, heutagogical and rhizomatic elements (Mendiola Medellín et al., 2023). Our contribution is not the claim that such hybridisation is possible, but its systematic anchoring in a single design apparatus: the andragogical learning contract as the device that selects and fixes, case by case, which principle governs each segment of the learning path, and a content-agnostic procedural map derived from problem solving in technology-rich environments and dynamised by double-loop learning. Where the PAH continuum arranges the three approaches along a developmental progression, our framework treats them as co-available design options governed by the contract; and where recent hybrid proposals assign andragogy, heutagogy and rhizomatic learning to distinct delivery formats (Mendiola Medellín et al., 2023), our framework unifies them through a single contract-governed procedure rather than mapping each to a separate mode.

Finally, the empirical basis of the framework is limited in ways that qualify its claims. The sixteen practices were purposively selected and documented by the project partners rather than sampled, and the underlying documentation is largely self-reported and therefore susceptible to positive reporting bias; the evidence is geographically concentrated in European Erasmus+ contexts; and no comparison of learner outcomes across practices was available. For these reasons, the practice analysis should be read as illustrative of the patterns that inform the framework, not as evidence of its effectiveness, which remains to be tested empirically. The framework is, moreover, culturally situated: it was developed in a European setting, and its relational and self-determination emphases may be understood and weighted differently within more communal, non-Western epistemologies—such as those that motivate proposals like ubuntugogy (Bangura, 2005)—so that transversality of design does not entail universality of outcome. Our characterisation of the proliferation of derivative paradigm labels as conceptual inflation concerns the analytical strategy of coining supposedly transcendent paradigms, not the substantive cultural insights that such concepts may carry.

8. Conclusions

This article has argued that the treatment of pedagogy, andragogy and heutagogy as rival, mutually exclusive paradigms has exhausted its productivity, and that the conditions of twenty-first-century adult learning are better served by a model that systematises their convergences. Therein lies its principal contribution: a shift from the comparison of paradigms to their integration. The ‘hybridised and harmonised’ andragogical framework proposed here rests on four assumptions: convergence over typified divergence; the continuity between self-direction and self-determination, both reducible to the learner’s freedom, autonomy and independence of judgement; the learning contract as the device within which principles from all three traditions can be legitimately combined; and hybridisation as the actual practice of adult education, finally given theoretical recognition. Its procedural translation—a modular, PS-TRE-inspired sequence framed by pre- and post-activity evaluation and dynamised by double-loop learning—makes the proposal implementable and, crucially, testable. The agenda it opens is twofold: empirically, the validation of the model in piloted training activities and the comparative study of contract-based hybrid designs; professionally, the formalisation and certification of the adult educator profile the model requires. Both are conditions for adult learning theory to stop reproducing its paradigm rivalry and start cumulating knowledge.

These conclusions carry implications for policy as much as for research. If the fragility and funding dependence that the practice analysis brought to light are to be overcome, the principles set out here cannot remain confined to isolated, project-based courses; they need to be embedded in the mainstream systems and policies of adult learning. Three priorities follow. First, adult education policy should incorporate hybrid, contract-based design into the standards that govern the provision, quality assurance and funding of adult training, so that the combination of pedagogical, andragogical and heutagogical principles becomes a criterion of ordinary provision rather than a feature of exceptional initiatives. Second, and correlatively, the adult educator profile the model presupposes should be formally recognised, professionalised and certified; the ten core competences defined within the project offer a concrete starting point for such qualification frameworks (Marcaletti et al., 2021). Third, provision should be organised along a life course logic, articulating age-specific strategies with stable mechanisms for validating and certifying both learners’ and trainers’ competences, so that continued learning is secured as a right across working life and beyond it, rather than left to the contingencies of short-term funding. Embedding the model at this systemic level is the condition under which its benefits can reach the adults—older, low-qualified and at greatest risk of exclusion—who stand to gain most from it.