MY youngest son was given an unusual alarm clock for Christmas, and over the vacation period he played with and became expert at adjusting this present. Having proved its trustworthiness in daytime trials he carefully set this electronic wonder to go off at 7am on his first day back at school. As expected, it started its lengthy musical verse at exactly 07.00 on the day. Hitting the snooze button instantly returned it to a quite state. A few minutes later the alarm started again, only this time with harsher and louder music. This process was repeated several times until my son exclaimed: "It is getting angry, I'd better get up." The emotional alarm clock has arrived.

While our emotional communication is complex and little understood, this alarm clock experience prompted some conjecture on the possibilities for emotion-sensing by technology in general. If machines could sense our state, what would they do, and could they improve our lot? My only direct experience has been the passive mouse and some interactive spacecraft manuals.

In the former case the worst possible tactile feedback comes into play when standing before an expectant audience, trying to line up a projector. If all goes well, no problem. But at the first hint of trouble, the hands start to sweat and the electrostatic coupling between finger and electronics starts to break down. This causes more tension and more sweat, and even worse coupling, and a rapidly worsening of this vital control function. A rapid downward spiral starts with ever more positive feedback. My only solution to date is to quickly wipe all moisture from my hands.

In the second case I was observing some adolescents voice-navigating a CD spacecraft manual, when one of them uttered a profanity. The machine responded sternly: "Do you really think you should address me in that way?" The impact was profound and the discourse that followed was noticeably more polite.

We often associate the word emotion with an unenviable, illogical, dangerous and unthinking human condition. However, the latest research seems to point to emotions being a vital ingredient in sensible decision-making. It turns out to be one of those valuable human qualities we often deride, but would miss if it were banished. This then begs the question; would machines and other artefacts benefit from emotions? If they could identify ours, and use them to communicate with us it might prove extremely valuable.

A very simple facility I have on my machines is the change of screen colour by memory state. With ample RAM available the screen is deep blue, but as RAM gets low, there is a transition to lighter shades of blue. When RAM becomes critical the it goes to pink, and it is time to take action.

What a simple but powerful icon, and what a pity there is no reciprocal path allowing machines to gauge my emotional state. If only they could slow down and speed up, bring in help files, and give me warnings as I make tension and fatigue-related errors.

Peter Cochrane holds the Collier Chair for the Public Understanding of Science & Technology at the University of Bristol. His home page is:
http://cochrane.org.uk