Skip to content

Computers and Science

April 29, 2010
tags: ,

A couple of weeks ago, Justin wrote about Eureqa, a computer program that can “discover” laws governing the behaviour of data input to it, expressing concern that computers may move us into a realm where we describe behaviour without having an understanding of it.

Last night at the Science Museum we saw a talk by Dame Jocelyn Bell-Burnell, the discoverer of pulsars. She had an interesting point about computers and their use in science.

She discovered pulsars by accident. The radio astronomy group she was working in was looking at quasars. The output data wasn’t in computer form (this was the late ’60s), it was on reams and reams of paper. Over the 6 month study they generated around 3.3 miles of paper plotting the observations. This all had to be analysed by hand. It was whilst looking at these miles of plots that the signals generated by pulsars were noticed, and most importantly, noticed as being unusual and different from what was expected. Her point here was that if a computer program had been used to analyse the data it would not have picked up these anomalies, as they were not in any way expected: when they were first noticed they were labelled “LGM”, or Little Green Men.

It’s an interesting point; if we use computer programs to analyse data and look for patterns, are they only looking for things that we might expect? Are they only looking in ways we already use? If so, are they going to be able to spot the things that move science on so well, the unexpected phenomena?

Advertisements
2 Comments leave one →
  1. Justin permalink
    April 30, 2010 4:33 AM

    Any statistical analysis done by humans is done just as well by computers. In particular, in order for these LGM to be noticeable they must deviate in some statistically significant way from some null hypothesis. A computer fitting the data according to some null hypothesis would necessarily report significant deviations.

    I don’t know the specifics of the discovery of pulsars, but to support her story and provide reason against the above let’s imagine a scenario:

    During months of pouring over data minor deviations were selected out by our Scientist’s eyes as being meaningful in some way, even though statistically they weren’t. It is very much in human nature to seek pattern even when there might not be (how relevant to this blog’s primary theme!). So these blips on a piece of paper might have been just like tea leaves settling in a fortune-teller’s cup or sticks thrown to select an I Ching reading. However, in the case of the dots on a piece of paper with some dots just recording noise and some dots recording quasar signal, our Scientist saw a sign — a signal if you wish. She then decides, contrary to rational evidence to tune down some channels and tune up others and suddenly the randomness fades as a true sign emerges. Our Scientist has gotten lucky. The tea leaves actually do have meaning.

    So what is it that humans can do that computers cannot do now? If we say there is no such thing as luck, that is our Scientist tried one possibility of many and happened to register a positive event, then Computers are far more lucky then humans. Computers were made to brute force billions of possibilities. So perhaps by lucky we mean our Scientist used intuition to sniff out the right result. In this case, the computers are defeated as they don’t have intuition as we think of it (Question: What is intuition anyway?), but this does not preclude future generations of computers from having it.

    It does seem like one limit on computers and AI is that we can only teach a computer to have our cognitive capabilities as soon as we understand that capability in ourselves (We can only teach intuition the moment we know what intuition is.). For those wishing the prolongation of time to the “singularity” one must wish that we are too stupid to understand ourselves and thus too stupid to design things that can in turn understand ourselves. This seems rather bleak to me, as I for one hope that we can eventually understand anything, but perhaps evolution has not built us to achieve our own ambitions.

  2. May 4, 2010 10:14 AM

    Sorry for the delay in replying, 4 days out on the water enjoying the sun…

    I think the point that I was trying to make here was that yes, intuition has a role to play (by intuition I mean “knowing” something without a (concious) rational process to arrive at this knowledge. In this case it would have been easy to dismiss the results as noise or a statistical “blip”; there was no rational explanation as to why the signals that were observed should exist. Bell-Burnell had to posit the existence of a new type of star to explain the observations, I’m questioning whether or not a computer would be able to do more than highlight the existence of a statistical anomaly.

    The role of the human mind here was to say that no, these were a significant result and then to derive an explanation for the data. Is this something a computer would be able to do? Come up with a new explanation (a new sort of star never before imagined) to explain data that could be described away if desired – this seems like a human thing to do. This obviously creates some issues in my mind, as I see the human mind as a deterministic-ish system, so I’ve got a bit of cognitive dissonance going on about these two ideas.

    A comment about your last sentence:

    This seems rather bleak to me, as I for one hope that we can eventually understand anything, but perhaps evolution has not built us to achieve our own ambitions.

    Can we ever know everything? This is something I was thinking about recently. We’re happy to accept that there are limits to the knowledge that a chimp, say, can have about the universe (i.e. less than humans). We’re happy to accept that a worm can have less knowledge than a chimp. The way I see it the natural thing to think is that there must be a limit to our potential knowledge. I don’t necessarily think that’s a bad thing; how will we ever know that there are things we don’t know? There are unknown unknowns. Ignorance is bliss!

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: