Following on from my last Blog, 20 years later… and are we really any further on?, I thought I’d try to add a bit more current detail. So, I went all 21st century and asked Google A.I. what it thinks (‘thinks’ – is that right? Ed) about the current body of Intensive Interaction research.
Well, despite me having some grave doubts about the current intelligence of Artificial Intelligence, Google A.I. said: ‘Research into the Intensive Interaction approach faces several limiting factors as of 2026, primarily centered on methodological rigor and the inherent nature of the participant group’ (accessed on 02/01/26).
So, in more detail, Google A.I. points to the following ‘Methodological Limitations‘ of the current body of Intensive Interaction research:
- Small Sample Sizes: Many studies rely on very small cohorts or individual case studies, which limits the ability to generalize findings to a broader population.
- Weak Research Designs: A significant portion of existing research uses AB designs (simple baseline and intervention phases) or semi-experimental methods rather than more rigorous randomized controlled trials.
- Baseline and Consistency Issues: Research is often hampered by unstable baselines or a total lack of baseline data before intervention starts. Additionally, there is often a lack of evidence regarding the long-term maintenance of effects.
- Variable Intervention Fidelity: It is frequently unclear if different practitioners are applying the approach consistently or correctly, making replication difficult.
So actually, I think most of this is a pretty accurate summation of the methodological limitations of much of the research into Intensive Interaction so far – although often there are good reasons for these limitations. However, it doesn’t mention the large number of these smaller cohort studies, and the cumulative confidence we should take from their combined, universally positive findings (which you can see in our downloadable Intensive Interaction Published Research Summaries Document).
Google A.I. also pointed to the following ‘Practical and Ethical Constraints‘:
- Complexity of Data: Measuring the outcomes of human interaction is difficult; some researchers argue that standard measurement tools can feel “reductionist” when applied to the fluid nature of Intensive Interaction.
- Participant Heterogeneity: The diverse needs of individuals with profound and multiple learning disabilities (PMLD) or autism mean that outcomes are often highly idiosyncratic and dependent on specific contextual factors.
- Ethical Hurdles: Conducting ethically sound research with this population is challenging, particularly regarding obtaining informed consent and ensuring that the individual remains the central focus.
- Limited High-Quality Publication: Historically, there has been a lack of peer-reviewed research, with many positive findings existing only as anecdotal reports or within “grey literature”*1.
Again, I actually think that most of this is fair comment – although I do have some issues here (and not just Google A.I.’s Americanized English spellings). Some of the points it makes are correct (IMHO – that’s ‘in my humble opinion’ for older readers) and succinctly explain how difficult robust empirical research can be in the area of human social interactivity. But some, I think, need to be looked at more closely.
One such issue, which has recently given me cause for thought, was raised in the paper ‘The Effectiveness of Intensive Interaction: A Systematic Literature Review‘ (Hutchinson & Bodicoat, 2015), which looked at 15 quantitative and 3 qualitative papers. The authors say: ‘Another issue when researching into interventions for people with PMID*2 is the heterogeneity of such a population group’ – as noted by Google A.I. above as ‘Participant Heterogeneity‘ – I suspect that it has read that paper!
So, it is this issue of wishing for greater ‘homogeneity’ within or amongst research cohorts that has somewhat irked me. The reason is that I think this issue is afforded too much significance as a research concern, thus weakening the case for Intensive Interaction to be judged a broadly effective approach based on the whole body of published research evidence. Yes, I think that too much analytical weight is, and historically has been, accorded to this issue … and here’s why I think that:
Despite wide differences in many personal attributes, characteristics and/or diagnoses of participants across Intensive Interaction research studies (i.e. their collective heterogeneity), I would say that they are ‘homogeneous’ regarding the most significant research variable when looking at a process of social inclusion (via Intensive Interaction) i.e. all the research participants have some degree of social or communication difficulties and/or differences, leading to their social exclusion when confronted by generally unempathetic social and communicative norms (due to ‘the double empathy problem‘: Milton, 2012).
Even if their physical, neurological or cognitive presentations vary quite widely, these differences, should I contend, be judged as individualistically negligible in analytic terms, i.e. they all encounter the same problem of social exclusion, as stated above, irrespective of their personal attributes, characteristics and/or diagnoses.
In conclusion, I think that the body of Intensive Interaction research should therefore be considered as much more widely ‘homogeneous’, and therefore more universally generalisable than has previously been accepted. Too much analytical weight has been given to other personally differentiating attributes that, in terms of social engagement or inclusion, are, I would say, to a large degree analytically irrelevant, i.e. in essence, Intensive Interaction should only be judged in terms of its efficacy in reference to the problem it is striving to address (Jones, O., 2026).
Anyway, Google A.I. goes on to say: ‘Ongoing efforts to address these factors include larger-scale projects like the INTERACT trial, a cluster-randomised superiority trial active through late 2025/2026, aimed at establishing clinical and cost-effectiveness for children with PMLD*3‘.
So, Google thinks things are on the up, and hopefully, it will be a much happier Google A.I. sometime in the near future.
As mentioned above, if you would like to see more details of the full body of published Intensive Interaction research, you can download the ‘Intensive Interaction Published Research Document 2025‘ at: https://connectingwithintensiveinteraction.com/wp-content/uploads/2025/03/the-intensive-interaction-published-research-summaries-document-2025.pdf
Notes:
*1 ‘Grey literature’ refers to research and information produced outside traditional commercial or academic publishing, such as government reports, conference papers, theses, white papers, and clinical trial data, often valuable for current insights but harder to find and typically lacking formal peer-review.
*2 PMID = Profound and Multiple Intellectual Disabilities.
*3 PMLD = Profound and Multiple Learning Difficulties.
References:
Google, A.I. (2026) ‘What are the limiting factors of research into Intensive Interaction approach?’ – accessed on 02/01/26.
Hutchinson, N. & Bodicoat, A. (2015) ‘The Effectiveness of Intensive Interaction: A Systematic Literature Review’, Journal of Applied Research in Intellectual Disabilities, 28: 437–454.
Jones, O. (2026) Personal communication.
Milton, D. (2012) ‘On the ontological status of autism: the ‘double empathy problem”, Disability & Society, 27(6): 883–887.
P.S. Many thanks to Oliver Jones for his invaluable feedback on this piece.
Connecting with Intensive Interaction 