Research and Evidence
An honest, measured summary of the current research evidence on social and companion robots in elderly and dementia care, written for commissioners, clinical leads and families who need to form a view without being misled by either hype or dismissal.
The research on social robots in elderly care is often cited in two contradictory ways: by technology advocates as proof that robots deliver measurable improvements in resident wellbeing, and by sceptics as thin, poorly controlled evidence that overpromises outcomes. Both characterisations contain some truth. This post tries to give a more accurate account of where the evidence is genuinely strong, where it is limited, and what it is reasonable to conclude.
This is not a systematic review. It is an editorial summary of the research landscape for commissioners, clinical leads and family members who need to form an informed view. Anyone making a procurement or clinical decision should review primary sources and seek specialist advice.
It helps to separate two categories of research that are often conflated. The first is evidence about therapeutic robots in dementia and palliative care settings, most prominently the PARO robotic seal developed at Japan's National Institute of Advanced Industrial Science and Technology. PARO has been the subject of a substantial body of peer-reviewed research over two decades, in multiple countries including the UK, and has a more developed evidence base than most other care robot devices.
The second is evidence about social and communication robots more broadly: devices designed to reduce loneliness, support social engagement and provide conversation for older people in care settings. This category includes screen-based systems like GenieConnect and AI conversational devices. The evidence base here is considerably thinner and more recent.
Treating these as a single category produces a misleading picture. The stronger evidence for PARO in dementia care does not automatically apply to communication robots used with cognitively intact older adults, and vice versa.
For PARO in particular, there is a reasonable body of peer-reviewed evidence suggesting benefit in specific circumstances. Systematic reviews and meta-analyses have found evidence of reduced agitation and anxiety in people living with dementia following interaction with PARO. Some studies have found improvements in mood and social engagement. Effects on depression have been less consistently demonstrated. No study has demonstrated clinically significant improvements in cognitive function, which is sometimes wrongly implied in advocacy materials.
The research in a UK context is more limited but consistent with international findings. UK studies have generally been small and conducted in specific dementia care settings rather than across the breadth of care home provision. The Alzheimer's Society has noted that robotic companions may be useful for some people living with dementia, while emphasising that individual variation is substantial and that the technology is not appropriate for everyone.
Importantly, research consistently finds that the effect is not uniform. Some individuals with dementia respond very positively to PARO and similar devices; others are indifferent or distressed by them. A care home deploying a therapeutic robot should not expect a reliable, predictable effect across all residents. Individual assessment and ongoing monitoring matter more than the device itself.
For more on this, see our guide to robots for dementia care in the UK.
The evidence base for communication and companion robots in residential care is less developed. Studies of devices designed to reduce loneliness and social isolation in older adults have produced mixed results. Some studies report that residents who interact regularly with a communication robot show improvements on loneliness scales or positive affect measures. Others find no significant effect or find that the effect disappears when the novelty of the technology wears off.
The methodological challenge is significant. Many studies lack adequate control conditions. It is genuinely difficult to establish whether an observed improvement in a resident's mood or social engagement is caused by the robot itself, or by the additional staff attention that introducing and monitoring the device generates. In a sector where staff time for social engagement is constrained, even the indirect effect of increased staff presence may matter; but this is not the same as the device producing the benefit.
There are also questions about durability. Short-term studies capture initial responses, which may be inflated by novelty. Studies that follow residents over a longer period are fewer in number and produce less consistent findings. This is not a reason to dismiss the technology, but it is a reason to be cautious about confident claims of sustained benefit.
Our guide on social robots and loneliness in elderly people covers the loneliness-specific evidence in more detail.
NHS-adjacent research into care robots has increased in recent years. NHS Trusts and Integrated Care Boards have conducted or supported pilots of companion technology as part of broader technology-enabled care strategies. These pilots are not always published in peer-reviewed form; many remain as internal evaluation reports.
The NHS Innovation Accelerator and AHSN (Academic Health Science Networks) have supported assistive technology projects that include social robot elements. The published evaluations from these programmes tend to be more cautious in their conclusions than vendor-produced case studies, which is a more accurate reflection of the evidence at this stage of the technology's development.
UK academic research centres, including groups at the universities of Lincoln, Hertfordshire and Plymouth, have published work on social robotics in care settings. The consistent finding across this body of work is that context, individual matching and integration into care practice matter more than the specific technology chosen. A well-chosen device introduced poorly is less beneficial than a less sophisticated device introduced with careful preparation and ongoing monitoring.
Several claims are regularly made about social robots in care settings that the current evidence does not support. Stating these plainly is part of responsible guidance in this sector.
The evidence does not support the claim that social robots can substitute for human social contact. No study has demonstrated that robot interaction produces equivalent outcomes to regular meaningful human interaction, and several have found that residents who have adequate human contact do not show additional benefit from robot interaction. The technology is supplementary, not substitutive.
The evidence does not support the claim that robots reduce care staff requirements. There is no research basis for the position that introducing a companion robot means a care home can operate with fewer care workers. This is an important distinction for commissioners who may be evaluating technology partly in response to workforce pressures.
The evidence does not support universal deployment. Research consistently finds that some residents benefit, some are indifferent and some are distressed. Deploying a device with all residents in a unit without individual assessment is not supported by the evidence and carries a risk of harm to those who respond negatively.
If you are a commissioner, occupational therapist or clinical lead evaluating social robot technology for a care setting, the evidence base supports a cautious, individual-led, evaluation-focused approach. It does not support a large-scale procurement on the basis of projected outcomes that the research does not reliably demonstrate.
A structured pilot with clear outcome measures, appropriate consent processes, and an honest evaluation against baseline data is the right framework. The evidence then produced from your own setting is more relevant to your procurement decision than any published study, because individual and organisational factors matter significantly.
If you would like to discuss a technology evaluation or pilot for your setting, write to hello@humanoidrobotcare.co.uk. We respond to all genuine professional enquiries.
Send an enquiryCommon questions
The evidence is mixed and more nuanced than either advocates or critics present. For therapeutic robots used with people living with dementia, there is a reasonable body of peer-reviewed evidence suggesting that some devices can reduce agitation and increase social engagement in certain individuals. The evidence for reducing loneliness in cognitively intact older adults is more limited and less consistent. Most studies are small and short-term. Evidence supports careful, consent-based use with appropriate individuals, not blanket deployment.
As of September 2026, NICE has not issued specific guidance on social or companion robots in care home settings. Care providers should review current NICE guidance on dementia care and assistive technology, and should not assume that the absence of specific NICE guidance either approves or disapproves the technology. Always check the NICE website for the most current guidance.
Most studies are small-scale, conducted over short periods and in conditions that may not reflect everyday care home use. Many lack adequate control conditions, making it difficult to separate the effect of the device from the effect of additional attention a research study generates. There is a recognised risk of publication bias toward positive findings. Many studies come from Japan or other Asian contexts that may not transfer directly to UK settings. These limitations mean strong claims of effectiveness are not well-supported, but they do not mean the technology is ineffective.