I’ve now finished most of week 0 – the exception being listening to the audio files, as these won’t be available until the course website opens on the 28th. I must remember to download and listen to them!
The rest of week 0 was fairly easy going and consisted of reading the start of the course guide, the preface and the first chapter of the main textbook (no chapter 0, I see …) and the introduction to the methods companion. I also successfully installed and activated my copies of SPSS (welcome back), E-Prime and the connectionist network software (ACT-R) we’ve been supplied with.
The course text for week 0 attempted to define cognitive psychology, put it into some kind of historical context, say a little about the cognitive approach by touching on the concepts of representation and computation and give an overview of Marr’s level-dependent explanations.
The part of the text that stuck out most for me however was the list of eight assumptions (based on Von Eckardt’s work) that the cognitive approach makes. These are:
1. Capacities can be partitioned and studied in isolation from one another (e.g memory vs language.)
2. Focus is on the individual and their environment – social /cultural aspects are de-emphasised.
3. Cognitive capacities are autonomous from non-cognitive ones – e.g. affect, motivation.
4. It is useful/meaningful to distinguish between normal and abnormal cognition.
5. It is possible to talk of a “typical” cognizer and generalise as adults are sufficiently alike to ignore any individual differences.
6. Answers to empirical questions can be given in terms of information processing.
7. Answers to empirical questions should be justified on empirical grounds.
8. Neuroscience constrains the possible answers to empirical questions.
I can see these making for interesting debate during the course and I haven’t even done the critical social psychology module, DD307, yet! But for the moment, I’m just glad to have got my study off to a reasonable start before the beginning of the working year on Monday.
As with ED209 I’m going to try to create notes as I go along to help with (a) getting my head around the material in the first place and (b) support my revision for the examination in October. If you’re also studying DD303 this year and find these notes helpful that’s great – but, please remember that I am a student too and this is my first time through the material as well! I’m also not likely to make many notes on topics that I (think I) understand well enough already – such as the early history of computing and Turing Test that appear in the first chapter as well as general experimental methodology and statistics. So caveat emptor, as some people say.
I do, however, welcome any constructive comments on them and on the blog in general. These were tremendously helpful to me during ED209 last year and it’s great to know that I’m not studying alone!
Good to see a fellow BSc MBCS (and CITP, fingers crossed!) on the course.
I was pondering, when I read the textbook section about the Turing Test, whether the test _as originally conceived_ allowed for any spontaneous initiation of topics by the computer. Any reports I’ve heard of the test actually being applied have talked about the computer’s response, as do your notes. The textbook, as far as I can see, doesn’t explicitly state whether the computer only responds to overtures from the tester rather than making any of its own. (I don’t count the lame responses which attempt to change the subject when the computer’s struggling!)
Do you happen to know whether Turing expressly specified responses? I wondered about it because voluntary action (like deciding to post a comment!) seems an important part of thinking.
P.S. Thanks for posting your notes!
Hi Gail,
Likewise, it’s nice to meet you too!
Turing’s original 1950 article, “Can Machines Think?” is available on the web (here, for example: http://www.loebner.net/Prizef/TuringArticle.html ) and as far as I can see, he definitely frames his test in the context of the machine giving answers, rather than asking novel questions. However, even Turing challenges his own question (see section 6 of the paper where he writes – “The original question, “Can machines think?” I believe to be too meaningless to deserve discussion.”)
The most interesting part of the paper (to me, anyway) is not his question and the test he devises, but its final paragraph. Here, Turing speculates on what might be possible in future and comes up with two suggestions. The first is to teach a machine to play chess, the second is to teach it, from first principles like a child, to understand English.
Sixty years on and the first has been achieved many times over. Most computer chess programs will consistently beat the average player and some will even beat the very best. Although the algorithms used are pretty sophisticated, there is still an inevitable amount of “brute force” in them – analysing many thousands of positions before deciding on the best strategy. It is manifestly not the case that human chess players operate in that way!
In the second endeavour, far less progress has been made. Perhaps this says something about what it is to “think” – and even the most sophisticated software doesn’t yet get close to this, in my view.
Tim.