As our population gets older and becomes more reliant on computer-based technology, we're going to need ways in which to not only store our memories but also to help those with memory loss - whether through accident, trauma or dementia.

The BCS Thought Leadership debate 'Aide Memoire: Prospects for Assistive Memory Technologies' held on 16 July, got together some of the leading figures from the academic, medical and business communities to discuss what we can do, what should be done and where the research challenges exist.

One of the traditional ways to help patients with memory loss is to use an aid of some sort. These are designed to assist people who need to take medication at certain times for example. Existing devices are fairly rudimentary and aren't much of a step forward from tying a knot in a handkerchief.

Some devices set off an alarm at the pre-set time to remind you about something, but don't tell you what that something is. Other devices, such as fan that as it spins displays a message on its blades, tell you what you should be doing but have no alarm. This makes them only useful if you happen to look at them. Little information and no sense of context are present in these devices.

As an example of the use of digital technology in developing new memory aids, Microsoft has developed a device called the SenseCam which is designed to record everyday events on a regular basis. Fitted with a fish-eye lens, it has no viewfinder but simply takes images at pre-determined intervals, not when the owner decides. As part of a funded experimental programme, these have been given to people who are suffering from memory loss.

In at least one instance using the SenseCam has helped a woman improve her memory by reviewing images the SenseCam has taken with her carer. By going over the images, and the events they document, the woman's memory was better than if she had used a diary - a more standard memory aid. Although promising, much more work is required before any general results can be deduced.

The use of devices such as the SenseCam in this area is still in its early stages, but in the future memory aid functionality could be integrated into devices that we already carry, such as mobile phones or watches that are linked using Bluetooth technology.

With hard disks doubling in capacity every 15-18 months, it was originally thought that we would store all our information on one huge disk. Although storage technology isn't yet capable of fitting 1000s of gigabytes of data on a laptop, when that day comes it will be theoretically possible to store a DVD quality movie of your life on it since one hour of DVD quality footage takes up about 1GB of storage space.

However, in recent years with the move to online storage different possibilities are emerging. With sites such as Flickr and Facebook and others now storing memories in lots of different ways, exactly how we will store our future memories is less predictable and could be highly distributed.

In fact not only does this trend mean that we will all have data in lots of different places, but it will also be in many formats. So not only do we need to decide where we are going to store what's most valuable to us, we need to make sure that it's done in a way that is accessible.

One thing that is certain, though, is that we need processes that integrate with our lives, otherwise we won't use them. We need intelligent software to act as a kind of companion to us that understands the situations that we're in - not just record them.

The challenge of building such companions is a variant of the challenge confronting Artificial Intelligence. How much adaptivity and intelligence do such companion system need before they become useful? Ideally such systems would be able to infer the desires, beliefs, and intuitions of their human users so as to offer up appropriate help, advice and control at the right moment and in the right format.

AI research has shown that humans are extraordinarily disposed to assume intelligence in devices - even if the behaviour is quite limited it only has to be appropriate to the situation. It is the human user who invests the response with intelligence. In order to build effective companions a slew of research challenges need to be progressed - ranging from perception through to communication, learning through to narrative generation.

Whilst explicit representations of memories via intelligence and rational behaviour is one aspect of effective support there are others that address more emotional and affective aspects of our sense of self.

For example, our appreciation and recall of music, rhythm and composition seem to be universal characteristics that are learnt extremely early and retained almost the end of life. Research is showing that musical memory may be exploited to change mood , enhance awareness and facilitate recollection.

There are many problems associated with using memory aid devices. Something that takes pictures could lead to the creation of the equivalent of the world's most boring holiday photos. Also, being reliant on devices could make us lazy in the way that use of calculators means many people can no longer do mental arithmetic.

On the flipside to this though, using a device to help us remember would free up our minds to remember other things. After all with the advent of calculators we no longer need to remember the algorithms for long division or for finding the square root of a number. Our capacity to spell may undergo a similar transformation in an age of automated spelling and grammar checking.

Another big issue is that of privacy. Storing all this information about yourself online is a huge security risk and represents a very different notion of what is held about an individual and for how long.

Then there's the issue of copyright. If someone posts an image of you on the internet or stores it in their personal files, who owns the copyright on your image? And ethically, who has the right to view your memory once it has been recorded?

Despite the fact that currently the amount of data being stored is outstripping the amount of storage space, the big issue for memory is how files will be indexed and then accessed.

Software works in a totally different way to the human mind, so it's not such a good idea to try and replicate the way we store memories. After all if you compare the way that we walk with the way that most forms of transport work, such as cars, it shows that different solutions can work in different contexts.

There's also the issue of whether people will want to wear these devices. Will it be socially obligatory for people to wear them? And if we're all wearing them, when is a memory yours and not a group one? Are they cultural or personal? Will there be a case for some memories being deleted if they are replicated by others, and who will decide?

As technology progresses electronics companies are likely to come out with hardware that can interact with our mental and cognitive states in order to store or augment our memories. Will these devices then also be allowed to alter our memories, delete them, as well as to keep them safe?

Even if all the advances in memory storage initially only benefit a small amount of people, there are bound to be spin-offs that will be more mass market.

One thing is certain as we get older  we all begin to lose our memories. If technology can enable us to store and subsequently retrieve our memories efficiently then it will benefit everyone, but we must be aware of the potential downsides and risks in providing such capabilities - not least the privacy, security and ethical issues that will arise as part of such developments.

August 2007