https://doi.org/10.66757/1988-592X-RE-2026-413-741
https://orcid.org/0000-0002-0770-7258
University of Malaga
The theory of multiple intelligences (MI), proposed by Howard Gardner in 1983, has had a significant impact on the educational field by challenging traditional conceptions of intelligence and promoting pedagogical approaches oriented toward diversity, inclusion, and individualized learning. Since its introduction, MI theory has been widely incorporated into school curricula, teacher education programs, and contemporary educational discourse. However, despite its popularity and practical influence, its status as a scientific theory remains highly contested. This article presents a critical theoretical review of MI that goes beyond the descriptive or pedagogical accounts commonly found in the literature. The analysis systematically examines its main conceptual, logical, and methodological weaknesses, as well as the implications of its uncritical adoption in educational contexts. In particular, the paper addresses the lack of robust empirical evidence supporting the existence of independent intelligences, the ambiguity surrounding the operational definition of “intelligence,” and the difficulties involved in developing valid and reliable measurement instruments. These issues are contrasted with alternative models of greater empirical strength, especially the theory of general intelligence (g factor), which is supported by extensive psychometric, neuroscientific, and predictive research. The article’s original contribution lies in linking these theoretical critiques to an analysis of the concrete pedagogical uses of MI in educational institutions and teacher training. By doing so, it questions the scientific claims often attributed to such practices and highlights the tension between pedagogical appeal and scientific validity. The paper argues that the continued prominence of MI in education is better explained by its flexibility and intuitive attractiveness than by its empirical rigor. On this basis, the article offers a reflection on the actual scope and limitations of MI in contemporary educational design.
Keywords: intelligence, education, multiple intelligences, psychology, methodology
The impact of Howard Gardner’s theory of multiple intelligences (hereinafter MIT) on education is undeniable. Since the publication of Frames of Mind in 1983, educational curricula and teacher training programs across much of the Western hemisphere have incorporated its principles to a greater or lesser extent. This influence has been particularly evident in the redefinition of student assessment criteria, the promotion of differentiated teaching strategies, and the legitimization of pedagogical practices aimed at personalizing learning. In school contexts characterized by cultural, cognitive, and socio-educational diversity, MIT has often been presented as a theoretical framework capable of responding to the demands of inclusion and educational equity.
However, the widespread dissemination of MIT in educational practice contrasts with the persistence of intense academic debate about its theoretical and empirical soundness. Numerous authors have pointed out conceptual shortcomings, logical problems, and methodological limitations that call into question its status as a scientific theory of intelligence. In response to these criticisms, Gardner and his followers have offered various counterarguments, with varying degrees of consistency and acceptance in specialized literature. In this context, it is pertinent to examine not only the internal coherence of MIT, but also the implications of its adoption as the theoretical basis for pedagogical and educational decisions at schools and faculties of education.
Based on this framework, this article aims to answer the following research question: is the theory of multiple intelligences sufficiently robust to justify the impact and centrality it has acquired in the education system? The hypothesis guiding this work study argues that it is not, insofar as MIT presents theoretical and empirical shortcomings that are difficult to overcome, which do not support its widespread presence or its use with scientific pretensions in educational practice.
To develop this argument, the article first sets out the main foundations of MIT and the positions of its advocates. It then critically analyzes its theoretical limitations and the available empirical evidence, paying particular attention to the studies cited in support of its educational application. Subsequently, it examines its influence in the field of teacher training and in contemporary pedagogical discourse, where its presence is constant despite the inconsistencies noted. Finally, the conclusions are presented, summarizing the line of argument followed and evaluating the validity of the hypothesis put forward.
The overall aim of this article is to critically study the theory of multiple intelligences (MIT) from a philosophical and epistemological perspective, evaluating its conceptual coherence, logical consistency and the soundness of the scientific assumptions that underpin its widespread adoption in the field of education. Based on this analysis, the aim is to assess whether the centrality that MIT has acquired in contemporary pedagogical discourse and teacher training is sufficiently justified from a theoretical and empirical point of view.
More specifically, the paper examines the concept of intelligence underlying MIT and its delimitation from related notions such as ability, aptitude, or talent; analyzes the main criticisms made from cognitive psychology, psychometrics, and neuroscience, with special attention to the absence of robust empirical evidence supporting the existence of independent intelligences; to contrast MIT with alternative models with greater empirical support, in particular the theory of general intelligence or g factor; and to reflect on the theoretical and practical effects of the uncritical adoption of MIT in educational contexts, especially in curriculum design and teacher training.
From a methodological point of view, the article adopts a qualitative approach of a theoretical-argumentative nature, typical of the philosophy of education and the epistemology of the social sciences. It is not an empirical or experimental research, but rather a critical theoretical review of the specialized literature, aimed at evaluating the conceptual and scientific validity of MIT and its suitability as a theoretical framework for educational practice.
The analysis is based on the review and comparison of relevant primary and secondary sources, including both the foundational works of Howard Gardner and the texts of his main advocates, as well as critical contributions from cognitive psychology, psychometrics, neuroscience, and the philosophy of science. The selection of sources is based on criteria of academic relevance, impact on scientific debate, and relevance to the analysis of the construct of intelligence.
The analytical procedure combines a logical-conceptual examination of the central assumptions and definitions of MIT, a contrastive argument with empirically supported alternative theories, and a critical evaluation of the theoretical justifications that support its educational application. This approach allows the work to be situated within the field of philosophical reflection applied to education, clearly differentiating it from empirical or evaluative studies, and contributes to clarifying the epistemological and practical limits of the analyzed theory.
In 1983, Howard Gardner proposed that the human mind is divided into seven or eight types of intelligences. Initially, his proposal achieved resounding success in the educational field in the United States and Canada; today, that success has also been replicated in educational curricula across Europe, with Spain being the case addressed in this article. For Perry D. Klein, Gardner’s proposal poses a dilemma: a weak version of the theory of multiple intelligences would be of little interest, he argues, whereas a strong version is not supported by the evidence presented by Gardner (Klein, 1997, p. 377). This criticism is echoed by various detractors of the theory, namely that there is insufficient scientific evidence to defend it unequivocally, which makes that it has become nothing less than a dogma in the organization of curricula an issue to consider.
In this regard, Klein points out another drawback related to the difficulties that are also presented by the general theory of intelligence, namely that it is too broad to plan a curriculum and, as a theory of abilities, too narrow, since it is static when it comes to assessing students’ competencies (Klein, 1997, p. 377). Other authors, such as Lynn Waterhouse (2006), have also argued that Gardner’s theory does not fit the empirical evidence, further suggesting that it is not consistent with the cognitive discoveries being made in the neurosciences.
For this reason, the author argued, it should not be applied in education, at least until these issues were remedied. This paper will return to these points, but first, it will present Gardner’s theory as the author himself introduces his arguments. Howard Gardner, now president of Project Zero, created in 1967 by Nelson Goodman at Harvard University, whose purpose was to study the development of artistic knowledge in children and adults (Gallego González, 2009), participated in this research as a postgraduate student.
His position regarding cognitive development in human beings changed considerably throughout his academic career. Initially, he shared with Piaget the theory of four stages of cognitive development: the sensorimotor period, the preoperational period, the period of concrete operations, and the period of formal operations. In the 1980s he took a different turn, as he sought to explain creativity, which in his view was ignored, along with biology, by Piaget, Wilhelm Stern’s IQ approach, and information processing, since their focus was placed on logico-linguistic problem solving (Gardner, 1993, p. 24).
Above all, for the author, intelligence entails the ability to solve problems or create products and tools that are valued within any community, a set of biopsychological potentials that are refined over time. In this way, these potentials, or intelligences, are not one, as the general intelligence theory assumes, but several, encompassing individuals’ talents, skills, and capacities. Everyone possesses these intelligences, although each person may be more prolific in some than in others or may find it easier to develop certain intelligences over others (Arnold and Fonseca, 2004).
It is relevant to clarify that Gardner does not deny the existence of a general factor of intelligence, nor does he enter into the debate concerning the heritability of intelligence (nature vs. nurture); however, he does exhibit a clear preference for environmental over genetic factors in cognitive development, with education playing a prominent role. In his words: “An intelligence is a biopsychological potential that is ours by virtue of our species membership. That potential can be realized to a greater or lesser extent as a consequence of the experiential, cultural, and motivational factors that affect a person” (Gardner, 1999, p. 82).
Frames of Mind (1983) is where Howard Gardner published his research on human intelligence, proposing that it is divided into up to eight different types, each operating in different cultural contexts, namely: linguistic, logico-mathematical, spatial, bodily-kinesthetic, musical, interpersonal, and intrapersonal. He later added what he chose to call “naturalistic intelligence,” which consists in understanding living beings and using that knowledge in practical ways, as a farmer or a fisherman does.
His methodology for identifying these intelligences consisted of applying eight criteria: 1) the study of damaged brain regions, 2) the existence of prodigies and geniuses, 3) the presence of one or more essential brain operations, 4) the existence of a definable set of actions indicating mastery of the skills, 5) the role of evolution, 6) susceptibility to encoding in a symbolic system, 7) support from psychological tasks revealing whether certain abilities are (or are not) manifestations of the same intelligences, and finally, 8) support from psychometric data (Gardner, 1993). In this list we see that, prima facie, there is an issue since the line distinguishing a skill or aptitude from an intelligence is not clearly drawn.
All intelligences have a specific set of operations and activities and are independent from one another; thus, some individuals will display characteristics of certain intelligences more than others (Gardner, 1983). To understand how this theory unfolds in its cognitive analysis, it is useful to examine how Gardner defines each of the intelligences he identifies, in addition to a final one he later suggests after the addendum of naturalist intelligence: existential intelligence.
Linguistic intelligence is characterized by lacking a concrete object or reference in the world. It would make use of both the right and left hemispheres of the brain, but is primarily located in the left temporal cortex, in Broca’s area (Morchio, 2004). It is related to fluency in oral and written language; in Morchio’s words, it is a skill “in the use of syntax, phonetics, semantics, and the pragmatic uses of language” (Morchio, 2004, p. 4). For Gardner, this is the intelligence that predominates in writers, lawyers, or speakers (Gardner, 1999).
Logical-mathematical intelligence is also located in the left hemisphere, although it would also employ the right hemisphere for the comprehension of numerical symbols. It entails the ability to use numbers effectively, in a scientific and logical manner (Gardner, 1999). According to the author, scientists, mathematicians, and logicians primarily exhibit this intelligence.
Spatial intelligence, by contrast, operates through the capacity to create two- and three-dimensional drawings, as well as to manipulate and understand spatial configurations (Gardner, 1999). For those in whom this intelligence predominates, recalling images of objects from memory is easier than recalling words. It is the intelligence of surgeons, painters, or architects.
Bodily-kinesthetic intelligence, in turn, entails the ability to use the body or parts of it to carry out activities, solve problems, or express ideas (Gardner, 1999). It is the intelligence associated with technical professions and sports (Gardner, 1999). For the author, intelligences that refer to objects in the world share a connection. In his words:
Bodily intelligence completes a trio of object-related intelligences: logical-mathematical intelligence, which grows out of the patterning of objects into numerical arrays; spatial intelligence, which focuses on the individual’s ability to transform objects within his environment and to make his way amidst a world of objects in space; and, bodily intelligence, which focusing inward, is limited to the exercise of one’s own body and, facing outward, entails physical actions on the objects in the world (Gardner, 1993, p. 235).
That is, these are intelligences that share a common space. On the one hand, objects are organized numerically, transformed, and used (or one’s own body is used as an object). It is important to recall that the types of intelligence are not entirely independent from one another; rather, depending on the individual, some are developed more than others, and sometimes several at once.
Interpersonal intelligence is related to assertiveness and the ability to assess others’ interests, motivations, intentions, or even to predict their actions or feelings (Gardner, 1993). Intrapersonal intelligence, by contrast, concerns self-knowledge: the capacity to view oneself from the outside and to explain which feelings or reasons guide one’s own actions (Gardner, 2006).
Near the end of the list appears the somewhat enigmatic naturalistic intelligence, which pertains to sensitivity toward natural, geological, or urban forms. Gardner describes it as the intelligence of a young person capable of discriminating among plants, birds, or dinosaurs (Gardner, 1999, p. 49). It is striking that in this description the author uses “intelligence” and “skill” interchangeably. Finally, he speculates about the existential intelligence, but does not fully consider it an intelligence because it fails to meet the criterion of evidential support. It would, in any case, pertain to questions concerning existence, life, death, or love (Gardner, 2006).
It is useful to introduce a conceptual clarification that allows for a coherent evaluation of the different presented models of intelligence. From a philosophical standpoint, a distinction may be drawn between intelligence as a general second-order capacity, understood as an explanatory principle that articulates the acquisition and coordination of abilities across various domains, and skills or aptitudes as first-order dispositions, observable and contextually specific. This distinction, common in the philosophy of mind and in the theory of scientific constructs, does not entail reifying intelligence, but it does require that intelligence fulfill a unifying explanatory function rather than a merely descriptive one.
From this perspective, the critique of the Theory of Multiple Intelligences is not directed at the diversity of human capacities, but at the expansion of the concept of intelligence to include what would more precisely correspond to skills or talents. At the same time, this distinction also compels an equivalent scrutiny of the theory of the g factor, acknowledging its status as a theoretical construct and its epistemological limitations. The relevant difference does not lie in ontological superiority, but in the degree of coherence among definition, empirical evidence, and explanatory power.
It is no surprise that, due to its success, the MIT immediately generated confrontation with the traditional school and with the previously imposed way of measuring intelligence, namely IQ tests. Gardner did not dismiss quantitative measures of intelligence in his proposal; rather, he maintained that they failed to capture the plurality of operations of human cognition, along with their theoretical limitation, however, his defenders did.
Among the most relevant criticisms of the theory of multiple intelligences is the absence of solid empirical support. Various authors have pointed out that the data provided by Gardner, who claimed to base his proposal on biological and neuropsychological studies, have not proven sufficiently robust or replicable in more controlled research. Consequently, there are no widely accepted psychometric tests that clearly validate the proposed classification of intelligences.
This criticism has been repeatedly formulated. In 1994, Sternberg stated that he was unable to identify any study that empirically validated Gardner’s theory, a conclusion corroborated by Alix in 2000 (Waterhouse, 2006, p. 208). That same year, Gardner himself acknowledged that there was very little strong empirical evidence supporting his approach (Gardner and Connell, 2000, p. 292). Later, Sternberg and Grigorenko insisted on this lack of support, to which Gardner responded by admitting that MIT had not been successful among psychologists who required psychometric or experimental evidence for validation (Waterhouse, 2006, p. 208).
In response to these objections, some defenders of the theory have argued that empirical methods do not constitute the sole criterion of scientific soundness and that, as a relatively recent proposal, it may not yet have adequate measurement instruments (Chen, 2004, p. 22). Moreover, it has been claimed that its validity rests on its purported successful application in educational contexts and on its better fit compared to traditional IQ tests.
However, applying a theory as the criterion of its own validation inverts the logical order between theory and practice. This position often draws on twentieth-century debates about the scientific method, appealing to the supposed impossibility of full methodological objectivity. Nevertheless, neither Karl Popper nor Thomas Kuhn maintained that empirical data were dispensable; on the contrary, Kuhn defended the need for experimental records to provide theories with a solid normative foundation (Fuller, 2004; Füller, 2004). The most frequently cited reference in this context is Paul Feyerabend’s Against Method (1975), where methodological dogmatism is criticized as a dead weight for science.
From this perspective, MIT has even been compared to historical episodes such as the case of Galileo, whose hypotheses were disseminated through effective rhetorical strategies before conclusive empirical proof was available, such as the famous experiment from the Tower of Pisa, which actually never took place. Following this logic, MIT would be validated by its application in classrooms. However, even granting this reasoning’s premises, Gardner maintained that the empirical basis of his theory lay in the positive educational results derived from its implementation (Gardner, 2004, p. 220), a stance that is logically problematic.
Indeed, practice cannot validate theory if its application presupposes its validity from the outset. Waterhouse further emphasizes that factors such as teacher enthusiasm or student motivation make it difficult to properly assess the effectiveness of educational interventions (Waterhouse, 2006, p. 208). It is also challenging to determine whether any positive effects are due to the theoretical framework as a whole or to isolated components independent of the proposed concept of intelligence.
Moreover, the criticism extends to the very definition of intelligence in MIT, where a shift occurs between intelligence and talent or skill. Intelligences such as musical or bodily-kinesthetic more accurately describe specific aptitudes than fundamental forms of intelligence. By contrast, the g factor model maintains the existence of an underlying general capacity that enables problem-solving across multiple domains.
Although defenders of MIT argue that intelligence is a construct defined by psychologists and not a tangible object (Chen, 2004), this does not preclude its measurement if its properties are adequately specified. Intangible constructs such as general intelligence have been successfully operationalized (Johnson and Bouchard, 2005). Gardner, however, acknowledged that he had been unable to define measurable characteristics of his intelligences due to their incompatibility with traditional psychometrics (Gardner, 2004, p. 215), which prevents empirical verification of the theory’s validity.
Thus, in order to analyze appropriately why the separation into multiple intelligences does not appear to function, it is relevant to return to the already mentioned g factor (general intelligence). Proponents of this approach argue that there exists a unitary intelligence identified by a “G” item, measured through IQ tests (McRorie and Cooper, 2004; Johnson and Bouchard, 2005). This intelligence seeks to reflect “brain efficiency” (Waterhouse, 2006, p. 2010).
In this case, empirical evidence does provide support: individual cognitive abilities are correlated with “G,” which can be used to predict intellectual performance under certain conditions and is interconnected with working memory abilities (Colom et al., 2004). The “G” theory therefore more accurately distinguishes between intelligence, defined by its functional role, and skills that would arise epiphenomenally. It also allows for the heritability of general intelligence to be reflected through functional magnetic resonance imaging that measures brain volume alongside the well-documented correlation between brain volume and performance on IQ tests (Toga and Thompson, 2005).
However, volume is not the only relevant factor. As Waterhouse notes from McDaniel’s (2005) work, the theory also enables the evaluation of supposed sex differences, citing a meta-analysis of 37 studies involving 1,530 men and women, showing that the correlation between IQ and brain volume is higher in women than in men. This finding is linked to the observation that women, despite having smaller brains, possess more gray matter. Understood in this way, intelligence may be more closely associated with brain activity (e.g., frontal lobe – IQ).
In Gardner’s work, a blurring of the boundaries between different intelligences can be observed. If these intelligences were truly independent, it would be difficult to explain the overlap apparent within his own theoretical framework. A clear example is the relationship between linguistic and logical-mathematical intelligence, since logical reasoning often requires linguistic precision, and vice versa.
This raises the question of whether these competencies are genuinely independent or whether they constitute facets of an underlying general intelligence. MIT theorists have not adequately specified what links the different intelligences, and Gardner himself acknowledged their semi-independent character, suggesting the possible existence of a “central agency of intelligences” (Gardner, 1983). This formulation is compatible with the idea of a general intelligence influencing performance across different domains.
These difficulties point to an ambiguity in the very definition of intelligence employed by MIT. No clear conceptual distinction is established between intelligence, capacity, and skill. By excessively broadening the concept, it loses operational clarity and explanatory precision. Furthermore, if eight intelligences are posited without a clear delimiting criterion, there are no grounds for excluding the existence of an indeterminate number of additional intelligences or internal subdivisions.
From a scientific perspective, a theory must allow for the formulation of clear predictions and be susceptible to measurement. In this respect, MIT has not demonstrated greater predictive capacity regarding performance in real-world situations than traditional models of intelligence. Attempts to develop assessment instruments based on multiple intelligences have been limited and have not achieved broad consensus.
Additionally, the theory relies heavily on anecdotal examples and biographical observations, such as the cases of Mozart or Einstein, which do not constitute sufficient evidence to demonstrate the existence of independent intelligences. Although evidence in favor of general intelligence does not necessarily exclude the possibility of multiple intelligences, it does highlight the difficulties of the model. Correlations among different abilities measured by IQ suggest that there is no entirely discrete processing of content “intelligence by intelligence,” as Gardner maintains (1999, p. 100).
Furthermore, the relationship between IQ, brain mass, and brain activity calls into question Gardner’s claim that these factors lack explanatory relevance. Indeed, his hypothesis of a “central agency of intelligences” linked to the frontal lobe is compatible with the g factor model, understood as executive function and working memory.
Taken together, the lack of empirical support and the internal conceptual inconsistencies, along with the robustness of findings derived from the theory of general intelligence, make a critical reevaluation of MIT necessary, particularly with regard to its educational application. Even appeals to its supposed practical success fail to resolve the logical problems of the model. It is noteworthy, finally, that Gardner himself later nuanced the scope of his proposal, partly acknowledging the provocative component that motivated its initial formulation.
On the one hand, the American psychologist did not anticipate such a profound impact of his theory on educational systems; initially, it was not designed for that purpose (Gardner, 2012). This does not mean that he did not consider its practical implementation. In his view, it is not appropriate to adapt every classroom concept to the eight intelligences, and the tendency to subject students to questionnaires to determine their type of intelligence incurred the same error that, in his opinion, IQ tests made. In that sense, a boy or girl could score very low in some intelligences, thereby introducing the very discriminations the theory sought to eliminate.
On the other hand, Gardner (2003) admitted that he used the term “intelligences” to problematize the issue of traditional schooling and intelligence testing. As he stated, had he used the term “skills” or “gifts,” the theory would not have attracted as much attention. He also added that he cannot claim the theory is correct or that its application will be successful, since further research is needed. In retrospect, it appears that concern about the theory’s empirical support dates to its origin but admitting that the reason the intelligences are not defined merely as skills lies partly in the term’s shock value telling: if accepted, it would render the theory more compatible with general intelligence or the “G” factor and would allow its postulates to be reevaluated in a different light.
It should be noted that the present analysis is framed as a critical theoretical review of the specialized literature on the theory of multiple intelligences. The selection of sources has been made according to their academic relevance and influence in the scientific and educational debate, prioritizing works by central figures in the discussion, such as Gardner, Sternberg, Waterhouse, and Johnson, as well as reviews and meta-analyses published in journals specializing in cognitive psychology, psychometrics, and educational sciences. The corpus analyzed includes both foundational texts and explicit defenses of MIT, as well as critical contributions evaluating its conceptual consistency and empirical support.
The scope of this review is not intended to be exhaustive in bibliographical terms, but rather analytical and representative of the main argumentative lines surrounding MIT and its contrast with alternative models of intelligence, particularly the g factor theory. This criterion distinguishes the present work from a purely argumentative essay, insofar as the theses defended are grounded in established debates and previously published empirical evidence, with the aim of assessing the scientific and educational relevance of MIT rather than proposing an alternative model of its own.
There is a significant gap between theory and practice, not only because multiple intelligences lack solid empirical support in classroom outcomes, as noted earlier, but also because their effective application is less widespread than often claimed. Although MIT appears in the curricula of many schools and in university-level teacher training, daily teaching practice reveals real difficulties in its implementation.
Gardner proposed ways of applying the theory, such as organizing the classroom into areas designed to explore materials linked to different types of intelligence (Gardner, 1991). In this way, teachers could identify and foster individual talents. He distinguishes between correct and incorrect ways of applying MIT, offering criteria for evaluating worthwhile pedagogical projects (Gardner, 1997). He has also collaborated with authors such as Chen on works concerning its practical implementation.
Studies on MIT’s impact on academic performance (Bas, 2016; Batdi, 2017) compare theory-inspired materials with traditional methods in science or language instruction. According to Ferrero, Vadillo, and León (2021), interventions based on MIT show superior results in experimental groups, although not always with statistically significant differences. They attribute discrepancies to methodological differences, such as sample size or the type of control group, noting that only Modirkhamene and Azhiri (2012) employed an active control group.
That study analyzed the reading comprehension of 70 secondary school students over two months using an MIT-based intervention, yielding positive results. However, Ferrero, Vadillo, and León highlight that it included practices considered invalid according to Gardner (1995), making it difficult to accept definitive conclusions. From these works, they conduct a critical review emphasizing the proliferation of MIT approaches and their theoretical and practical shortcomings. The principal studies on MIT’s impact on academic achievement are Bas (2016) and Batdi (2017). According to Ferrero, Vadillo, and León, both suffer from methodological deficiencies that limit their value in justifying educational application. The former is based on master’s theses in Turkey (1998–2014), and the latter on 63 articles and doctoral dissertations (2000–2016) from various countries (Ferrero et al., 2021, p. 2).
In this regard, both studies applied only minimal quality criteria in selecting studies and did not attempt to assess the risk of bias introduced by their procedures. Due to these shortcomings, the two meta-analyses reported very large effects (d = 1.077 and 0.95, as noted by Ferrero, Vadillo, and León). With such limited evidence and such large effects, it is doubtful whether these can truly be attributed to MIT-based interventions or, alternatively, to biases in the meta-analyses. Moreover, the search criteria for primary sources are insufficiently defined to allow replication, and no comprehensive list of included and excluded studies is provided to confirm or extend the purported findings.
For this article, it is therefore relevant to present the study conducted by Ferrero, Vadillo, and León, which builds upon the results of Bas and Batdi, as well as their subsequent reflections on the implications for implementing MIT. The authors clearly define their methodology, beginning with search criteria. Their task was to locate both studies (Bas, 2016; Batdi, 2017) in journals indexed in Web of Science and accessible through ProQuest and Google Scholar, using a defined and replicable procedure. Another objective was to assess the quality of each study, identify sources of bias, and measure their impact.
Regarding method, they specify that they followed the reporting standards for systematic reviews and meta-analyses recommended by PRISMA (Moher et al., 2009) and the APA. For the analysis, Marta Ferrero conducted a search limited to English-language articles published after the release of Frames of Mind (1983), using the term “multiple intelligences,” yielding 937 studies in Web of Science, 1,642 in ProQuest, and 944 in Google Scholar (Ferrero et al., 2021, p. 6). The inclusion criteria were: (a) quantitatively measuring the impact on students’ academic performance based on interventions using MIT, and (b) employing a pretest-posttest design with a control group.
This resulted in 38 articles in the first search and 232 in the second, which were independently read by the authors and screened according to the inclusion criteria, resulting in a selection of 15 studies. From these, 129 additional studies were identified by examining cited works, screened in the same way, and reduced to 26 in the first search and 13 in the second. The final sample thus consisted of 39 studies (Ferrero et al., 2021, p. 6). As noted, the primary contribution of this work was to assess the quality of the publications and identify potential biases.
Qualitatively, the results showed that the selected studies had serious methodological flaws and failed to provide sufficient information on essential elements necessary for critical evaluation, for example, whether participants were free from experimental manipulation or whether the measures used were reliable. There were also too few studies allowing replication, rendering MIT-based interventions something of a mystery. Even when information was available, studies often failed to meet quality standards.
The data analysis replicated previous meta-analytic findings, including the similarly large effect sizes reported in Bas (2016) and Batdi (2017). Compared with effect sizes typically found in psychological research (Funder and Ozer, 2019; Rubio-Aparicio et al., 2018), there is considerable disparity. Ferrero and colleagues attribute this to the least precise studies, those with the smallest sample sizes, such that the predominance of low-quality studies inflates overall results (Ferrero et al., 2021, p. 12).
Furthermore, while other areas of psychological research are required to meet strict methodological standards, studies supporting MIT rarely comply with such standards, and only one employed an active control group. As shown in the review, pedagogical studies relying on passive control groups tend to produce inflated effects, whereas the efficacy of educational interventions is better assessed using active controls. In sum, these shortcomings, along with the lack of preregistration (a safeguard against bias and cherry-picking) and errors in measurement methods, call into question the theoretical and practical validity of MIT.
It is nevertheless important to qualify the interpretation of the reported effect sizes in these meta-analyses to avoid an overly simplistic reading of their magnitude. In educational intervention research, large effect sizes are not in themselves unusual, especially when interventions involve comprehensive pedagogical changes, relatively short implementation periods, and high initial variability in student performance (Hattie, 2009; Hill, Bloom, Black, and Lipsey, 2008). From this perspective, values such as d = 1.077 or d = 0.95 may be plausible in certain educational contexts and should not automatically be dismissed by direct comparison with laboratory-based psychological studies or general observational research.
However, the central issue lies not merely in the absolute magnitude of effect sizes, but in the methodological conditions under which they are estimated. As Ferrero, Vadillo, and León (2021) show, the studies reporting the largest effects often coincide with lower methodological quality, small samples, absence of active control groups, and insufficient descriptions of interventions. Educational methodology literature consistently documents that studies using passive control groups tend to produce inflated effect estimates compared to those employing active or comparable alternative interventions, due to factors such as the Hawthorne effect, teacher expectations, or lack of control over concurrent pedagogical variables (Cheung and Slavin, 2016; Kraft, 2020). Thus, the critique is not directed at the theoretical possibility of large effects in education, but at the difficulty of confidently attributing such effects to MIT specifically when studies fail to meet minimal standards of methodological rigor and transparency.
Therefore, although large effect sizes may be coherent with certain educational interventions, the overall analysis of available evidence suggests that, in the case of MIT, such magnitudes must be interpreted cautiously. The combination of very large effects with a predominance of low-precision, high-risk-of-bias studies weakens their explanatory force and reinforces the thesis that the purported empirical effectiveness of MIT may be overstated. This conclusion does not a priori invalidate every intervention inspired by the theory, but it seriously questions its use as a scientifically consolidated theoretical framework for guiding educational practice with claims of demonstrated effectiveness.
As Ferrero, Vadillo, and León note, it is necessary to remember that the existence of multiple intelligences is not validated by the scientific community, which, as discussed in this article, tends to favor a g factor or general intelligence. Moreover, school applications yield mixed results, which ultimately do not serve to validate the theory’s practical claim that students should be developed according to their intelligence types. The persistence of such applications without empirical support poses significant risks for educational practice, including the adoption of ineffective interventions, misallocation of resources, and pedagogical decision-making based on untested assumptions. Available empirical evidence instead suggests that learning is largely determined by the nature of the material being learned (Willingham, 2004), rather than by learners’ supposed intelligence profiles (Ferrero et al., 2021, p. 12).
The theory has helped address certain shortcomings of traditional schooling, such as discrimination based on learning speed or overemphasis on mathematical and linguistic abilities at the expense of music or kinesthesia. However, replacing one reductionist model with another conceptually weak one risks perpetuating new forms of labeling under the guise of inclusion. While this may explain the enthusiasm it has generated among educators and the perceived positive effects in terms of motivation or talent recognition, it does not provide demonstrated criteria of positive impact sufficient to justify systematic implementation.
Such justification cannot be self-grounded and requires proven methods, though this does not imply rejecting alternative approaches sensitive to student diversity. More robust frameworks do exist, such as evidence-based differentiated instruction, Universal Design for Learning (UDL), or cognitive models supported by psychometrics, which address individual differences without relying on unvalidated intelligence categories. In this sense, a complementary approach is viable only if it is recognized that the concept of intelligence employed by MIT is inappropriate, as it adds no substantive information beyond that already available from general intelligence theory.
Despite these deficiencies, it has become commonplace in pedagogical theory and educational dissemination to treat MIT as valid, disregarding or ignoring available empirical evidence. This increases the risk that widely disseminated educational practices are legitimized by social consensus or intuitive appeal rather than demonstrated effectiveness. As briefly analyzed below, numerous popular and academic studies continue to assume MIT’s validity without prior critical evaluation.
At the popular level, one example is the review published by “la Caixa” Foundation of Multiple Intelligences Around the World (Chen, Moran, and Gardner, 2009) and Grit: The Power of Passion and Perseverance (Duckworth, 2016). Entitled “Toward a More Inclusive Education: From Multiple Intelligences to the Passion for Learning,” the review acknowledges certain practical limitations, such as financial investment or greater applicability in favorable socioeconomic contexts, but does not sufficiently address the epistemological and pedagogical risks of applying an unvalidated theory as a foundation for educational policy.
In the academic sphere, numerous studies employ MIT as an explanatory framework without subjecting it to critical evaluation. For instance, research interrelating creativity, gender, and age with multiple intelligences uses quasi-experimental designs to draw educational conclusions, assuming the validity of the theory that ought to be under scrutiny (Ramírez, Prado, and López, 2018). Similar patterns occur in studies that begin from the supposed “axiom” of MIT, even while acknowledging a lack of consensus in neuroscience and psychometrics (Lázaro-Tortosa et al., 2021).
Other works emphasize its potential for inclusion and talent identification (Díaz-Posada et al., 2017), explaining its broad dissemination in countries such as Spain between 2010 and 2020, both in teacher training and educational projects. Nevertheless, such widespread use without rigorous evaluation risks consolidating practices whose real effectiveness has not been demonstrated, thereby displacing better-founded alternatives.
In conclusion, although some works underscore the importance of fostering individual aptitudes, the relatively uncritical acceptance of MIT and the presumed benefits of its application constitutes a risk for evidence-based educational practice. Attention to individual differences does not require adopting a multiple intelligences model and can instead be articulated through more robust cognitive theories aligned with psychometric and neuroscientific advances. Ignoring this distinction perpetuates the gap between scientific research and educational practice that this work invites readers to reconsider critically.
This article has critically examined the Multiple Intelligences Theory (MIT) proposed by Howard Gardner and its sustained influence in the field of education since its formulation in 1983. Although MIT represented a break with traditional approaches by challenging the primacy of a general intelligence and by promoting pedagogical practices oriented toward diversity and the personalization of learning, the analysis conducted here highlights significant theoretical and empirical weaknesses that undermine its solidity as a scientific theory.
In particular, MIT has been shown to suffer from conceptual problems concerning the definition and delimitation of the construct of intelligence, as well as from a lack of robust empirical evidence supporting the existence of independent intelligences. The defenses put forward by Gardner and his followers, centered on the heuristic and inclusive value of the theory, are insufficient to address the most substantial criticisms, especially with regard to its scientific validation.
Despite these limitations, MIT continues to occupy a central place in educational curricula and teacher training, largely due to its practical appeal and ease of implementation across diverse pedagogical contexts. However, the findings of this study suggest that such prominence is not fully justified from a theoretical or empirical standpoint, and that its persistence owes more to pragmatic considerations than to scientific rigor.
Consequently, the validity and relevance of MIT in the contemporary debate on intelligence and learning should be subject to renewed critical scrutiny. The hypothesis advanced at the outset of this study is thus supported: Gardner’s theory does not justify its prevalence within the educational system under scientific pretensions. Without denying its possible utility as a limited pedagogical resource, its widespread adoption should be reconsidered in favor of a more rigorous and evidence-based educational debate.
Allix, N. M. (2000). The Theory of Multiple Intelligences: A Case of Missing Cognitive Matter. Australian Journal of Education, 44, 272–88. DOI: https://doi.org/10.1177/000494410004400306.
Andreu, M. (2017). Un colegio pionero en implantar la teoría de las inteligencias múltiples, Información, 12 de julio de 2017. Available at: https://www.informacion.es/elche/2017/07/12/colegio-pionero-implantar-teoria-inteligencias-5941746.html.
Arnold, J. and Fonseca, M. C. (2009). Multiple Intelligence Theory and Foreign Language Learning: A Brain-Based Perspective. International Journal of English Studies, 4(1), 119–36.
Bas, G. (2016). The Effect of Multiple Intelligences Theory–Based Education on Academic Achievement: A Meta–Analytic Review. Educational Sciences: Theory and Practice, 16, 183–1864. DOI: 10.12738/estp.2016.6.0015.
Batdi, V. (2017). The effect of multiple intelligences on academic achievement: A meta–analytic and thematic study. Educational Sciences: Theory and Practice, 17, 2057–2092. DOI: 10.12738/estp.2017.6.0104.
Chen, J. (2004). Theory of Multiple Intelligence: Is It a Scientific Theory? Teacher College Record, 106, 17–23. DOI: https://doi.org/10.1111/j.1467-9620.2004.0031.
Chen, J. Q., Moran, S., and Gardner, H. (2009). Multiple Intelligences Around the World. Nueva York: John Wiley & Sons. DOI: 10.3390/jintelligence6030038.
Díaz-Posada, L.-E., Varela-Londoño, S.-P., and Rodríguez-Burgos, L.-P. (2017). Inteligencias múltiples e implementación del currículo: avances, tendencias y oportunidades. Revista de psicodidáctica, 22(1), 69–83. DOI: https://doi.org/10.1016/S1136-1034(17)30046-1.
Ferrero, M., Vadillo, M. A. and León, S. P. (2021). A Valid Evaluation of The Theory of Multiple Intelligences is Not Yet Possible: Problems of Methodological Quality for Intervention Studies, Intelligence, 99, 1–14. DOI: 10.1016/j.intell.2021.101566.
Feyerabend, P (1975). Against Method: Outline of an Anarchistic Theory of Knowledge. London: New Left Books.
Fuller, S. (2004). Kuhn vs. Popper: The Struggle for the Soul of Science. New York: Columbia University Press.
Funder, D. C., & Ozer, D. J. (2019). Evaluating effect size in psychological research: Sense and nonsense. Advances in Methods and Practices in Psychological Science, 2, 156–68. DOI: https://doi.org/10.1177/2515245919847202.
Gallego González, Sofía (2009). La teoría de las inteligencias múltiples en la enseñanza-aprendizaje de español como lengua extranjera. Doctoral dissertation, University of Salamanca. DOI: 10.14201/gredos.76442.
Gardner, H. (1983). Frames of Mind: The Theory of Multiple Intelligences. New York: Basic Books.
Gardner, H. (1999). Intelligence Reframed: Multiple Intelligences for the 21st Century. Nueva York: Basic Books.
Gardner, H., & Connell, M. (2000). Response to Nicholas Allix. Australian Journal of Education, 44(3), 288–93. DOI: https://doi.org/10.1177/000494410004400307.
Hattie, J. (2009). Visible learning: A synthesis of over 800 meta-analyses relating to achievement. Routledge.
Hill, C. J., Bloom, H. S., Black, A. R., and Lipsey, M. W. (2008). Empirical benchmarks for interpreting effect sizes in research. Child Development Perspectives, 2(3), 172–77. https://doi.org/10.1111/j.1750-8606.2008.00061.x
Johnson, W. and Bouchard, T. J., Jr. (2005). The Structure of Human Intelligence: It is Verbal, Perceptual, and Image Rotation (VPR), Not Fluid and Crystallized. Intelligence, 33, 393–416. DOI: https://doi.org/10.1016/j.intell.2004.12.002.
Klein, P.D. (1997). Multiplying the Problems of Intelligence by Eight: A Critique of Gardner’s Theory. Canadian Journal of Education, 22, 377. DOI: https://doi.org/10.2307/1585790.
Kraft, M. A. (2018). Interpreting effect sizes of education interventions. Educational Researcher, 47(4), 241–53. DOI: https://doi.org/10.3102/0013189X18765750.
Lázaro-Tortosa, F. R., Marco-Arenas, M., and Sánchez-López, M. C. (2021). Identificación del talento musical en escolares de la Comunidad Autónoma de la Región de Murcia en España. Revista Educare, 25(3), 1–15. DOI: http://dx.doi.org/10.15359/ree.25-3.17.
McDaniel, M. A. (2005). Big-Brained People Are Smarter: A Meta-Analysis of The Relationship Between in Vivo Brain Volume and Intelligence. Intelligence, 33, 337–46.
McRorie, M., & Cooper, C. (2004). Synaptic Transmission Correlates of General Mental Ability. Intelligence, 32, 263–75. DOI: 10.1016/j.intell.2004.11.005.
Modirkhamene, S., & Azhiri, M. H. B. (2012). The Effect of Multiple Intelligences–Based Reading Tasks on EFL Learners’ Reading Comprehension. Theory and Practice in Language Studies, 2, 1013. DOI: 10.4304/tpls.2.5.1013-1021.
Moher, D., Liberati, A., Tetzlaff, J., & Altman, D. G. (2009). Preferred Reporting Items for Systematic Reviews and Meta-analyses: The PRISMA statement. BMJ, 339, b2535.
Morchio M. (2004). Enseñanza de una lengua extranjera desde las inteligencias múltiples. Córdoba, España: Universidad Nacional de Córdoba, Programa Universitario de Adultos Mayores. DOI: https://doi.org/10.47865/igob.vol6.n21.2023.235.
Rubio-Aparicio, M., Marín-Martínez, F., Sánchez-Meca, J., and López-López, J. A. (2018). A Methodological Review of Meta–Analyses of The Effectiveness of Clinical Psychology Treatments. Behavior Research Methods, 50, 2057–2073. DOI: 10.3758/s13428-017-0973-8.
Toga, A. W., & Thompson, P. M. (2005). Genetics Of Brain Structure and Intelligence. Annual Review of Neuroscience, 28, 1–23. DOI: 10.1146/annurev.neuro.28.061604.135655.
Waterhouse, L. (2006). Multiple Intelligences, the Mozart Effect, and Emotional Intelligence: A Critical Review. Educational Psychologist, 41(4), 207–25. DOI: 10.1207/s15326985ep4104_1.
Willingham, D. T. (2021). Why Don’t Students Like School? A Cognitive Scientist Answers Questions About How the Mind Works and What It Means for The Classroom. San Francisco: Jossey–Bass.