Why multiple intelligences theory is a neuromyth
Waterhouse L · 2023
grade Dcritiqueindependentnot-applicablenumbers spot-checked
Sample
Conceptual analysis of 40 years of evidence bearing on MI theory; includes two literature counts run in January 2023 ('multiple intelligences classroom' 1983-2022 returned 407 papers in Education Full Text and 92 in PubMed; 'validity Gardner multiple intelligences' returned 9 and 2 respectively)
Population
Classroom practice worldwide
Design
Published as a Frontiers 'Conceptual Analysis' - an argued critique with database counts, not a systematic review; no protocol, no pooling, hence grade D. The current statement of the case against MI, and unusually explicit about which of MI's claims have been tested and which have not. Notes Gardner's own defence - that MI cannot be a neuromyth because he never called it a neurological theory - and answers it by quoting Gardner asserting each intelligence "clearly involves processes that are carried out by dedicated neural networks" (Gardner 2020a p.94). Waterhouse has argued this case since 2006, so the paper is a restatement by a long-standing critic rather than an independent adjudication.
Key findings
No researcher has ever directly looked for a brain basis for the individual intelligences - the two papers claiming otherwise (Shearer & Karanian 2017, reviewing 318 brain-region studies; Shearer 2020b, reviewing 48 connectivity studies) are alignment exercises that never measure brain activity during behaviour characteristic of an intelligence, and Shearer's "unique neural architecture" for linguistic intelligence comprises eight regions amounting to nearly the whole brain. Factor studies have not shown the intelligences to be independent (Visser 2006, Almeida 2010, Castejon 2010). Studies of MI teaching effects have not explored alternative causes for positive effects (repetition, novelty, extra individual attention, teacher enthusiasm) nor been conducted to standard scientific methods. There are still no standard measures of the intelligences, so MI studies cannot be compared or synthesised at all; Gardner has declined to build them. Over 40 years neuroscience has shown the brain is not organised into separate modules dedicated to specific forms of cognition - Anderson 2013's meta-analysis of 2,000+ imaging studies found functional diversity at every location. Argues that belief in MI crowds out the search for evidence-based teaching methods - the opportunity-cost argument. Gardner himself has written that the "intelligences are fictions - at most, useful fictions".
Genetic confound
Not applicable; conceptual critique.
Effects
| Outcome | Metric | Value | Measure | Timing | Vs | Horizon | Class |
|---|---|---|---|---|---|---|---|
| Direct neuroscientific tests of the individual intelligences | count of published validity studies | zero. A January 2023 search on "validity Gardner multiple intelligences" (1983-2022) returned 2 papers in PubMed and 9 in Education Full Text; only one (Shearer 2020b) even attempted a neural-validity assessment, and it is an alignment review, not a measurement study | unknown | 1983-2022 | none | not-applicable | g |
| Independence of the intelligences in factor studies | qualitative verdict | not shown - Visser 2006 (200 adults), Almeida 2010 (294 children, CFA) and Castejon 2010 all found the intelligences intercorrelated; Plucker 1996 is the lone contrary result | standardized | n/a | none | not-applicable | g |
| Standard psychometric measures of the eight intelligences | count | none exist. Gardner et al. 1998's measures never became standard and Gardner has declined to build them; the one widely used instrument (Teele Inventory, used in 1,000 schools) was found by McMahon 2004 not to provide consistent measurement. Without shared measures MI findings cannot be compared or synthesised | standardized | n/a | none | not-applicable | g |
| Classroom effectiveness of MI teaching (the half of the literature that cuts AGAINST a dismissive verdict) | direction of the primary literature and of its meta-analyses | mostly positive. Waterhouse concedes that "most of the published classroom studies found improved learning through use of MI teaching methods" and that all three meta-analyses of MI teaching (Bas 2016, Batdi 2017, Ferrero 2021) report many positive results. Her argument is confounding, not nullity: Ferrero 2021 found most reviewed studies lacked adequate sample sizes, appropriate control groups or valid outcome measures, and none controlled for repetition, novelty, extra teacher attention or teacher enthusiasm. This is why the topic verdict is insufficient rather than no-effect | mixed | 1983-2022 | business-as-usual | end-of-treatment | domain-skill |
| Prevalence of MI belief and use among teachers | percentage | 90% of surveyed US teacher trainees planned to use MI teaching strategies (Ruhaak & Cook 2018); 94% of surveyed Quebec teachers said they used MI in the classroom (Blanchette Sarrasin 2019); only 25% of teachers surveyed in Canada, the US and the UK thought MI was a neuromyth (Craig 2021) | self-report survey | 2018-2021 | none | not-applicable | non-cognitive |
Cited by
- Should Gardner's multiple intelligences be used as a basis for instruction?insufficientconf: mediumgc: medium