In all your dealings with elderly relatives with dementia, it is vital to bear in mind that they retain memories of feelings, even though their capacity to retain recent facts is almost nonexistent. Alas, you will probably need to explain this to any "experts" into whose hands they fall.

There is a tragic refusal on the part of the psychiatric and psychological professions, and the Alzheimer's Society to engage with this simple fact. The Nice guidelines still advocate methods that can only cause deep distress.

Imagine you were unable to retain facts from more than 30 seconds ago. Suppose I was to wait a minute after you had finished this article and give you a test on its contents. How do you think it would make you feel to realise that, not only could you not recall a single fact in the article, you did not even know you had read it?

Yet Nice guidelines insist that practitioners must take a detailed history of the patient and conduct a test of their cognitive and emotional state. This must be accompanied by seeking consent to do the examination, and an explanation of the options. No one who understands what it is like not to retain new information could possibly suggest such a thing.

The likely consequence is to prove to them that something is grievously, terrifyingly wrong with their capacity for fact retention. No problem, the professional may reply, as the patient does not remember, they will soon stop worrying. Except that good evidence now exists that upsetting someone with dementia is disastrous because they do retain emotions.

Researchers showed people with amnesia film clips with a sad theme. Afterwards, they were still sad for some time, although often having no knowledge that they have even seen a film, let alone anything about its content. For example, one of the subjects cried during the films, and her face displaying sadness throughout. Shortly after it finished, she could not recall a single detail of what had been in the film. Yet her sadness lingered for 30 minutes.

Interestingly, exactly the same result was obtained when they were exposed to films with happy themes. Although unable to explain why, they felt happy afterwards.

The authors stress these findings' crucial implications for the care of people with dementia. For example, staff need to be trained to understand that if a patient seems unhappy for no apparent reason, it could be related to something that had happened some time before. Perhaps a phone call from a relative or a visit has upset them, but they do not realise that this is why.

Indeed, the implications of this and related research go far wider than dementia. If we can retain emotions unrelated to the facts that elicited them, that could explain why we seem unaccountably depressed at times. Our childhood and subsequent experiences live on in us. If something triggers the emotion without the associated facts (perhaps because repressed), that could explain much depression.

So addicted are the medical establishment to genetic causes of emotional states, and to the drug company's solutions, that they close their eyes and ears to anything that might challenge their model. Never mind that the Human Genome Project is demonstrating that hardly any of our vulnerability to mental illness is caused by genes.

In the case of depression, let's just keep on doling out the pills or cognitive behavioural therapy rather than getting to the cause of the problem. In the case of dementia, let's keep on asking them toxic questions that only serve to derange and terrify.

Emotion experiment: Feinstein, JS et al, 2009, Proceedings of the National Academy of Science, vol 107, No 17, 7674-9. More Oliver James at selfishcapitalist.com