Skip to main content Skip to navigation


<?xml version="1.0"?>

<!DOCTYPE TEI.2 SYSTEM "base.dtd">





<publicationStmt><distributor>BASE and Oxford Text Archive</distributor>


<availability><p>The British Academic Spoken English (BASE) corpus was developed at the

Universities of Warwick and Reading, under the directorship of Hilary Nesi

(Centre for English Language Teacher Education, Warwick) and Paul Thompson

(Department of Applied Linguistics, Reading), with funding from BALEAP,

EURALEX, the British Academy and the Arts and Humanities Research Board. The

original recordings are held at the Universities of Warwick and Reading, and

at the Oxford Text Archive and may be consulted by bona fide researchers

upon written application to any of the holding bodies.

The BASE corpus is freely available to researchers who agree to the

following conditions:</p>

<p>1. The recordings and transcriptions should not be modified in any


<p>2. The recordings and transcriptions should be used for research purposes

only; they should not be reproduced in teaching materials</p>

<p>3. The recordings and transcriptions should not be reproduced in full for

a wider audience/readership, although researchers are free to quote short

passages of text (up to 200 running words from any given speech event)</p>

<p>4. The corpus developers should be informed of all presentations or

publications arising from analysis of the corpus</p><p>

Researchers should acknowledge their use of the corpus using the following

form of words:

The recordings and transcriptions used in this study come from the British

Academic Spoken English (BASE) corpus, which was developed at the

Universities of Warwick and Reading under the directorship of Hilary Nesi

(Warwick) and Paul Thompson (Reading). Corpus development was assisted by

funding from the Universities of Warwick and Reading, BALEAP, EURALEX, the

British Academy and the Arts and Humanities Research Board. </p></availability>




<recording dur="00:50:06" n="6223">


<respStmt><name>BASE team</name>



<langUsage><language id="en">English</language>



<person id="nm1208" role="main speaker" n="n" sex="m"><p>nm1208, main speaker, non-student, male</p></person>

<person id="sf1209" role="participant" n="s" sex="f"><p>sf1209, participant, student, female</p></person>

<person id="sf1210" role="participant" n="s" sex="f"><p>sf1210, participant, student, female</p></person>

<person id="sf1211" role="participant" n="s" sex="f"><p>sf1211, participant, student, female</p></person>

<person id="sf1212" role="participant" n="s" sex="f"><p>sf1212, participant, student, female</p></person>

<personGrp id="ss" role="audience" size="m"><p>ss, audience, medium group </p></personGrp>

<personGrp id="sl" role="all" size="m"><p>sl, all, medium group</p></personGrp>

<personGrp role="speakers" size="7"><p>number of speakers: 7</p></personGrp>





<item n="speechevent">Lecture</item>

<item n="acaddept">Psychology</item>

<item n="acaddiv">ss</item>

<item n="partlevel">UG2</item>

<item n="module">unknown</item>




<u who="nm1208"> # <pause dur="1.9"/> so first thing i'll say is that # from <pause dur="0.3"/> # beginning of next week Monday <pause dur="0.4"/> # these <pause dur="0.2"/> # lectures will be moving into brand spanking new <pause dur="0.5"/> sort of flashy lecture theatre <pause dur="0.6"/> # in the new agriculture building <pause dur="0.3"/> and # <pause dur="1.5"/> this is # where it is <pause dur="0.5"/> so # <reading>approach the front entry in meteorology</reading> sort of right <pause dur="0.3"/> down there <pause dur="0.3"/> # <pause dur="2.5"/> <reading>go <trunc>outsi</trunc> go round outside left <shift feature="voice" new="laugh"/>side<shift feature="voice" new="normal"/> of meteorology to meet in reach the entrance to the new <trunc>ag</trunc> agriculture building <pause dur="0.5"/> lectures <pause dur="0.4"/> theatres one and two are on a lower ground floor <pause dur="1.1"/> use stairs or the lift <pause dur="2.3"/> and there will be no access to the new building from the old <pause dur="0.2"/> agriculture</reading> <pause dur="2.2"/> right <pause dur="0.4"/> does that <pause dur="0.3"/> make sense to everybody <pause dur="1.5"/> cool <pause dur="1.2"/> right <pause dur="0.2"/> # what i'm going to talk about today is expertise <pause dur="1.7"/> and # the first thing <pause dur="0.2"/> we need to do <pause dur="0.5"/> before <pause dur="0.4"/> we embark on an investigation of any sort of area <pause dur="0.7"/> is to <pause dur="0.2"/> define <pause dur="0.9"/> what we're talking about so that we know that we're actually talking <pause dur="0.4"/> about the same sort of thing <pause dur="1.7"/> so <pause dur="0.3"/> what constitutes <pause dur="0.2"/> an expert <pause dur="6.3"/> <kinesic desc="changes transparency" iterated="y" dur="6"/> right <pause dur="0.2"/> here's # one example from <pause dur="0.2"/> New Forest Post ninth of March <pause dur="0.3"/> nineteen-ninety-five <pause dur="1.0"/> #

front page article driving death off the road <pause dur="0.4"/> <reading>young drivers at a Forest School have been undergoing psychological tests that can help slash <pause dur="0.4"/> the death toll on Britain's roads an expert <pause dur="0.2"/> from <gap reason="name" extent="1 word"/> University <pause dur="0.3"/> has been putting forty-five Brockenhurst College students through hazard perception tests this week <pause dur="0.7"/> a hazard perception test could well be part of the driving test in two years time said Dr <gap reason="name" extent="1 word"/> in nineteen-ninety-five</reading> <vocal desc="laughter" n="ss" iterated="y" dur="2"/> <pause dur="2.3"/> <event desc="takes off transparency" iterated="n"/> never mind <shift feature="voice" new="laugh"/> # <vocal desc="laughter" n="sl" iterated="y" dur="1"/> so in <shift feature="voice" new="normal"/> this case this journalist from New Forest Post <pause dur="0.4"/> has labelled <pause dur="0.3"/> me an expert because somehow i managed to convince him <pause dur="0.5"/> that i knew marginally more about the subject area <pause dur="0.2"/> than he did <pause dur="1.0"/> however what does it say <pause dur="0.5"/> in the dictionary <pause dur="8.0"/><kinesic desc="puts on transparency" iterated="n"/><gap reason="inaudible" extent="1 sec"/><pause dur="2.5"/> okay it's a <pause dur="1.2"/> bog standard # Oxford Dictionary <pause dur="2.6"/> # <pause dur="1.0"/> an expert means <pause dur="0.2"/> being practised skilful and <pause dur="0.2"/> well informed <pause dur="1.1"/> and obviously the sort of problem with this <pause dur="0.4"/> definition is how <pause dur="0.3"/> exactly practised skilful and well informed do you actually have to be <pause dur="0.3"/> before we start calling you <pause dur="0.4"/> an expert <pause dur="1.5"/> and #

one area where the definition <pause dur="0.2"/> of expertise is <pause dur="1.6"/> vital is within the justice system <pause dur="0.6"/> so # expertise can actually be officially <pause dur="0.4"/> # conferred <pause dur="0.2"/> by a judge <pause dur="0.4"/> for the purpose of court room testimony <pause dur="0.6"/> and in fact <trunc>f</trunc> quite a few of my colleagues have been called up to court to give evidence as <pause dur="0.3"/> psychologists on various <pause dur="0.5"/> sort of different aspects <pause dur="0.2"/> particularly driving <trunc>dri</trunc> driver fatigue <pause dur="0.5"/> and various things to do with driving <pause dur="1.3"/> # <pause dur="2.5"/> and in this case <pause dur="0.8"/> the # <pause dur="0.2"/> definition of expertise <pause dur="0.5"/> is usually <pause dur="0.2"/> one on training this is # <pause dur="1.7"/><kinesic desc="reveals covered part of transparency" iterated="n"/> an official one this is actually in your handout <pause dur="6.1"/> okay so # for the purposes of justice <reading>the courts describe an expert as one who's qualified <pause dur="0.6"/> by training or experience to form a definitive opinion <pause dur="0.6"/> in areas of science arts or department of trade <pause dur="0.4"/> in which laypersons <pause dur="0.3"/> are not <pause dur="0.3"/> prepared to reach accurate conclusions</reading> <pause dur="2.0"/> but the problem with this sort of definition <pause dur="0.2"/> is that there is evidence that training <pause dur="0.7"/> and experience <pause dur="0.3"/> are <pause dur="0.2"/> poor criterion for judging the quality of an expert <pause dur="0.4"/> something we'll

<pause dur="0.4"/> explore <pause dur="0.6"/> # later <pause dur="2.9"/> so # <pause dur="0.5"/> moving on to <pause dur="0.2"/> more psychological <pause dur="0.4"/><event desc="takes off transparency" iterated="n"/> like definitions <pause dur="0.9"/> <reading>expertise has been seen as the amount and complexity of knowledge <pause dur="0.6"/> gained through experience in a domain</reading> <pause dur="0.6"/> okay by the way this # <pause dur="0.5"/> these # <pause dur="0.2"/> definitions are in your handout so i think they are yeah they should be in your handout so <pause dur="0.3"/> don't need to scribble them down <pause dur="1.3"/> that's all right <pause dur="2.0"/> okay this is good old Eysenck and Keane <kinesic desc="puts on transparency" iterated="n"/><pause dur="1.0"/> our course textbook <pause dur="1.7"/> and # they stated that expertise means being good at specific <pause dur="0.2"/> problems <pause dur="0.7"/> within a specific domain <pause dur="1.4"/> in a lot of domains <pause dur="0.2"/> expertise <pause dur="0.2"/> is defined by peer review <pause dur="1.4"/> for example with driving <pause dur="0.2"/> top # police officers <pause dur="0.5"/> # top police drivers are often put forward <pause dur="0.4"/> as <pause dur="0.3"/> the model of driving <pause dur="0.4"/> so they can perform <pause dur="0.3"/> a range of skills <pause dur="0.6"/> which have <pause dur="0.2"/> been defined <pause dur="0.2"/> by other <pause dur="0.5"/> police drivers <pause dur="0.5"/> as being important <pause dur="0.2"/> to good driving <pause dur="0.3"/> like # skid control hazard perception <pause dur="0.5"/> vehicle control et cetera <pause dur="1.6"/> and most of the time this may be entirely sensible <pause dur="0.8"/> so <pause dur="0.3"/> you can assume that they must know something about what they're

doing <pause dur="1.4"/> however police drivers have got a job to do <pause dur="0.6"/> and that job involves driving <pause dur="0.2"/> as fast as possible <pause dur="0.2"/> from A <pause dur="0.4"/> to <pause dur="0.2"/> B <pause dur="2.5"/> and with some <pause dur="0.2"/> advanced driving courses <pause dur="0.5"/> # which are usually run <pause dur="0.3"/> by <pause dur="0.2"/> police drivers <pause dur="0.8"/> # <pause dur="0.3"/> one problem with this is that people are often taught to drive <pause dur="0.2"/> like policemen <pause dur="1.0"/> and while this includes skills <pause dur="0.3"/> which have been shown to be beneficial to reducing accident involvement <pause dur="0.4"/> like hazard perception <pause dur="1.3"/> it can also legitimize <pause dur="0.2"/> practices <pause dur="0.3"/> which are inappropriate for people <pause dur="0.2"/> who aren't <pause dur="0.7"/> driving police cars <pause dur="0.5"/> so <pause dur="0.2"/> things like driving fast or making progress <pause dur="3.5"/> i'm very bitter about that 'cause i recently failed my motorbike test for failing to make progress by driving too slow round roundabouts <pause dur="0.7"/><vocal desc="laughter" n="sl" iterated="y" dur="2"/> and # <pause dur="3.0"/> here's another definition put forward by <kinesic desc="reveals covered part of transparency" iterated="n"/> some psychologists <pause dur="3.0"/> # <pause dur="0.4"/> Ericsson and Lehmann nineteen-ninety-six <pause dur="1.8"/> argued that # <pause dur="0.4"/> <reading>expertise expert performance involves <pause dur="0.2"/> an individual <pause dur="0.8"/> being able to produce a <pause dur="0.3"/> consistent performance <pause dur="0.3"/> on demand <pause dur="0.5"/> over a range of tasks that are representative of

of a domain</reading> <pause dur="0.9"/> that is <pause dur="0.3"/> expert <trunc>perf</trunc> # expert performers <pause dur="0.5"/> are those individuals who can reliably <pause dur="0.2"/> reproduce <pause dur="0.8"/> their skill <pause dur="0.2"/> their superior skill whenever <pause dur="0.3"/> you ask them to <pause dur="2.9"/> Ericsson and Lehmann <pause dur="0.2"/> argued that though this is quite a narrow definition <pause dur="0.3"/> the advantage <pause dur="0.4"/> of this sort of definition <pause dur="0.6"/> is that at last we have an opportunity <pause dur="0.5"/> for measuring <pause dur="0.3"/> expertise <pause dur="0.6"/> okay what we're doing here is operationalizing expertise <pause dur="0.5"/> so <pause dur="0.3"/> by by definition that means <pause dur="0.4"/> we can start measuring it in laboratory <pause dur="0.3"/> experiments the problem with all the previous <pause dur="0.3"/> definitions <pause dur="0.2"/> is because they're so nicely woolly and vague <pause dur="0.4"/> it's very hard to actually sit down and measure what we mean <pause dur="0.8"/> # <pause dur="0.2"/> by narrowing down the definition like this <pause dur="0.2"/> we can actually start measuring it <pause dur="0.4"/> which means we can start doing some science on it <pause dur="3.7"/> but as you can imagine <pause dur="0.5"/> evaluating <pause dur="0.2"/> performance is a real <pause dur="0.2"/> problem <pause dur="0.2"/> for this <pause dur="0.2"/> field <pause dur="4.1"/> for example what's a <pause dur="0.3"/> laboratory test for whether someone is an expert art <pause dur="0.2"/> critic <pause dur="1.3"/> so we could <pause dur="0.3"/> do something like <pause dur="0.5"/> develop a test of

their knowledge <pause dur="0.5"/> of history of art but would that <pause dur="0.2"/> be a fair test of whether they were a good critic or not <pause dur="2.2"/> so there's great difficulty in <trunc>operi</trunc> <trunc>operisati</trunc> # <pause dur="0.2"/> operationalizing <pause dur="0.3"/> expertise <pause dur="0.5"/> in many domains <pause dur="0.2"/> so <pause dur="0.3"/> and actually measuring expertise <pause dur="0.2"/> quantitatively <pause dur="1.3"/> and <pause dur="0.4"/> what we find with our police drivers' peer review <pause dur="0.5"/> is often the only <pause dur="0.5"/> measure of success <pause dur="1.1"/> and # <pause dur="0.6"/> Goodyear in nineteen-ninety-seven <pause dur="0.5"/> brought up this point with respect to professional psychologists <pause dur="0.8"/> so with professional counselling <pause dur="0.2"/> psychologists for example <pause dur="0.5"/> how can we actually measure the quality of their work you know is it the number of people they successfully treat in the year <pause dur="0.3"/> in a year or something like that <pause dur="1.0"/> and <pause dur="0.3"/> with academic psychologists like me is it the number of papers published the number of research contracts i've won <pause dur="0.5"/> not the quality of the papers published <pause dur="0.4"/> okay <pause dur="0.2"/> and these are all things that <pause dur="0.2"/> you know are actually used to evaluate us <pause dur="0.4"/> in <trunc>ra</trunc> real life <pause dur="0.7"/> but you know <pause dur="0.2"/> are they <pause dur="0.2"/> fair measures <pause dur="2.5"/> okay <pause dur="0.3"/> i mean and

you can see there's potential problems for that so for example <pause dur="0.4"/> # <pause dur="0.5"/> # we've we get although we get judged for teaching <pause dur="0.4"/> on a sort of department wide basis <pause dur="0.3"/> the teaching quality isn't actually <pause dur="0.2"/> formally assessed on a person by person basis <pause dur="0.2"/> okay apart from <pause dur="0.4"/> we get informal bits of feedback from the sort of sheets you fill in <pause dur="0.4"/> but # <pause dur="0.3"/> the actual formal assessment takes place on a department wide <trunc>proc</trunc> # process <pause dur="0.4"/> so at the moment <pause dur="0.4"/> # i actually have no personal <trunc>s</trunc> incentive for my career to <pause dur="0.2"/> sort of <pause dur="0.5"/> bother to sort of do any good teaching <pause dur="0.3"/> my all my incentive is based on research because that's <pause dur="0.2"/> what i personally get judged on <pause dur="0.2"/> not the teaching <pause dur="1.4"/> okay so there are problems <pause dur="0.3"/> in judging expertise <pause dur="0.3"/> in this sort of way <pause dur="1.6"/> so <pause dur="0.7"/> into the science how have people <pause dur="0.2"/> tried to get around <pause dur="0.5"/> these sort of problems <pause dur="1.7"/> well first thing they did <pause dur="0.3"/> was to choose <pause dur="0.2"/> a <pause dur="0.2"/> an appropriate domain <pause dur="0.7"/> and they chose a domain <pause dur="0.4"/> where expertise is well defined <pause dur="0.6"/> easy to operationalize <pause dur="0.4"/> and <pause dur="0.4"/> measurable <pause dur="1.1"/> and that domain <pause dur="0.9"/><kinesic desc="changes transparency" iterated="y" dur="11"/> was <pause dur="0.2"/> chess <pause dur="10.0"/> okay <pause dur="0.3"/> so now <pause dur="0.4"/> chess was the # <pause dur="0.2"/>

original domain for studying <pause dur="0.2"/> expert <pause dur="0.2"/> performance <pause dur="2.0"/> and the nice thing about chess is that there are <pause dur="0.2"/> clear criterion <pause dur="0.4"/> distinguishing novices <pause dur="0.2"/> from experts basically experts will beat <pause dur="0.5"/> novices <pause dur="1.2"/> and it's also <pause dur="0.2"/> a very well defined game set within a very rigid <pause dur="0.2"/> framework of rules <pause dur="0.8"/> and this makes it <pause dur="0.2"/> really nice for <pause dur="0.2"/> experimental work <pause dur="1.6"/> on the face of it it should be straightforward <pause dur="0.4"/> okay there's no <pause dur="0.3"/> # chance involved there's no <trunc>do</trunc> dice rolling or anything that that <pause dur="0.6"/> and there are a finite <pause dur="0.3"/> in theory finite set of moves that can be made at any one time <pause dur="1.5"/> but it's the sheer number <pause dur="0.2"/> of <pause dur="0.3"/> these possible moves which is overwhelming <pause dur="0.6"/> so just <pause dur="0.3"/> three moves into a chess game <pause dur="0.5"/> gives you more than nine-million <pause dur="1.7"/> distinct <pause dur="0.3"/> board positions <pause dur="5.4"/> and # on the face of it <pause dur="0.3"/> this should be an ideal task for a computer <pause dur="3.0"/> however attempts to design computers that can match <pause dur="0.3"/> the best chess players <pause dur="0.3"/> hasn't proved to be easy <pause dur="1.0"/> so <pause dur="0.7"/> what # computer chess <trunc>brai</trunc> # programs tend to do <pause dur="0.3"/> is search <pause dur="0.4"/> all available options <pause dur="1.1"/> for example a

a recent one <pause dur="0.4"/> developed by I-B-M <pause dur="0.5"/> and known as Deep Blue <pause dur="2.6"/> considered about ninety-billion <pause dur="0.6"/> moves <pause dur="0.4"/> per turn <pause dur="2.1"/> at a rate of about nine <pause dur="0.2"/> billion <pause dur="0.7"/> per second <pause dur="5.0"/> okay and this is a computer that i think it eventually managed to beat Gary Kasparov but it was a close run thing <pause dur="1.5"/> but the point is <pause dur="0.7"/> that people <pause dur="0.2"/> can't do that <pause dur="0.8"/> our brains literally can't think <pause dur="0.3"/> at anywhere near that sort of speed <pause dur="1.4"/> so how come the best <pause dur="0.2"/> chess players human chess players still manage <pause dur="0.2"/> to beat computers most <pause dur="0.4"/> of the time <pause dur="1.3"/> well what are the best chess players doing <pause dur="0.5"/> that the computers <pause dur="0.2"/> and the novice chess players aren't doing <pause dur="3.0"/> originally <pause dur="1.7"/> it was assumed that chess experts were better <pause dur="0.4"/> as a result of superior intellectual capacity <pause dur="3.0"/> however <pause dur="0.2"/> it's # since been shown that expert chess players <pause dur="0.3"/> aren't actually any <pause dur="0.3"/> cleverer <pause dur="0.5"/> than anyone <pause dur="0.2"/> else or <pause dur="0.2"/> cleverer than a matched control group i should say <pause dur="0.9"/> so <pause dur="0.2"/> study found that their I-Qs are no higher <pause dur="0.2"/> than matched <pause dur="0.3"/> controls <pause dur="5.5"/> so <pause dur="0.7"/> if their mental capacity <pause dur="0.8"/> doesn't appear to be any higher <pause dur="0.2"/>

than normal <pause dur="1.1"/> then presumably it must be their strategies <pause dur="0.8"/> which are superior <pause dur="4.2"/> so <pause dur="0.5"/> how has a chess performance been measured <pause dur="1.2"/> # lot of the <pause dur="0.6"/> original work which was carried out <pause dur="0.6"/> by <pause dur="0.3"/> de Groot <pause dur="1.7"/> and # <pause dur="0.4"/> what he did was get # <pause dur="0.2"/> two groups of chess players <pause dur="1.9"/> the first group were what he called experts <pause dur="0.3"/> and they were <pause dur="0.3"/> top local club players <pause dur="1.3"/> that's his definition of expert <pause dur="0.7"/> and he compared with an <pause dur="1.0"/> with an even better group <pause dur="1.1"/> the grand masters <pause dur="0.9"/> so <pause dur="0.2"/> grand masters are world class chess players <pause dur="0.2"/> so his comparison is between <pause dur="0.7"/> one really high level of expertise and a <pause dur="0.3"/> still a pretty high level of expertise and what <pause dur="0.2"/> he wanted to find out what is the difference <pause dur="0.2"/> between <pause dur="0.2"/> these two <pause dur="0.4"/> groups <pause dur="2.3"/><kinesic desc="reveals covered part of transparency" iterated="n"/> and what he did was # measure it measure <pause dur="0.5"/> # their performance and what they're thinking about <pause dur="0.5"/> with # <pause dur="0.2"/> verbal protocols <pause dur="0.6"/> and what verbal protocols are is simply getting people to talk aloud <pause dur="0.3"/> when they're doing the task <pause dur="0.2"/> and <pause dur="0.2"/> then transcribing what they're saying and trying to analyse it <pause dur="1.1"/> okay so he analysed their performance by

getting them to talk aloud <pause dur="0.4"/> and he also timed <pause dur="0.6"/> their responses <pause dur="4.1"/> and # he presented them with a <pause dur="0.3"/> game <pause dur="0.3"/> position <pause dur="1.5"/> and asked them to <pause dur="0.5"/> think aloud <pause dur="0.5"/> while deciding <pause dur="0.6"/> what move to make <pause dur="5.5"/> and <pause dur="1.0"/> he found that people were only considering <pause dur="0.3"/> about thirty <pause dur="0.3"/> alternative <pause dur="0.2"/> moves <pause dur="0.6"/> searching at most <pause dur="0.5"/> to a depth <pause dur="0.2"/> of six moves <pause dur="0.7"/> and <pause dur="0.3"/> frequently less <pause dur="3.3"/> okay <pause dur="0.2"/> and there's a <pause dur="0.2"/> nice summary of all this sort of stuff in Eysenck and Keane <pause dur="5.2"/> so the grand masters didn't consider more moves <pause dur="0.4"/> and they didn't search <pause dur="0.3"/> more deeply than the experts <pause dur="0.6"/> but they were <pause dur="0.3"/> <trunc>neverlethe</trunc> nevertheless slightly faster <pause dur="0.5"/> at getting to the right <pause dur="0.3"/> move <pause dur="2.0"/> and they also came to their chosen move <pause dur="0.2"/> earlier <pause dur="0.3"/> in their search through possible alternatives <pause dur="5.1"/> and # <pause dur="0.4"/> that was supported by the finding that the quality of the chess moves that these people were making <pause dur="0.7"/> remained high even when you <pause dur="0.2"/> forcibly reduced their search time <pause dur="0.4"/> okay so the <pause dur="0.3"/> the best players <pause dur="0.4"/> # still searching similar number of options <pause dur="0.3"/> but they're coming to the right option earlier <pause dur="0.4"/> in their search <pause dur="4.8"/> and in <trunc>a</trunc> in

addition <pause dur="0.2"/> they found that the grand masters <pause dur="0.7"/> # actually came up with better moves <pause dur="0.4"/> which was assessed by independent <pause dur="0.7"/> raters <pause dur="5.2"/> and what # de Groot argued here <pause dur="0.7"/> is that the chess master doesn't generate moves by <pause dur="0.4"/> a simple search <pause dur="0.5"/> through possible alternatives <pause dur="0.7"/> which is exactly what our chess computers are doing <pause dur="0.2"/> to generate all possible <trunc>altern</trunc> alternatives and then go through them <pause dur="0.4"/> one by one <pause dur="0.9"/> instead our grand masters <pause dur="0.3"/> are using <pause dur="0.7"/> cued <pause dur="0.2"/> recall <pause dur="0.4"/> from memory <pause dur="11.4"/> okay so as i said the depth of search how many moves people are thinking ahead <pause dur="0.5"/> # increases up the level of our experts <pause dur="0.5"/> but it doesn't actually increase when people move on to be <pause dur="0.2"/> grand masters <pause dur="3.0"/> and de Groot suggests that the difference between these experts and these grand masters is that the grand masters have a <pause dur="0.2"/> superior knowledge <pause dur="0.7"/> of different <pause dur="0.2"/> board positions <pause dur="0.6"/> and what the best move from each position would be <pause dur="8.6"/> okay so the idea is that the expert <pause dur="0.2"/> can recognize <pause dur="0.6"/> the position <pause dur="0.7"/> and <pause dur="0.3"/> knows a good move from there <pause dur="0.5"/> and that means you don't

have to think about <pause dur="0.2"/> all of the irrelevant moves <pause dur="0.4"/> which you know in advance are a load of rubbish <pause dur="0.4"/> because <pause dur="0.4"/> you've already you can recognize this pattern from your memory <pause dur="0.5"/> so you don't have to do what the computer is doing and go though all of the <pause dur="0.4"/> irrelevant <pause dur="0.3"/> <trunc>s</trunc> # possible moves as well <pause dur="0.7"/> and <trunc>ma</trunc> that makes it much more efficient <pause dur="0.4"/> and it means that you don't have to do <pause dur="0.5"/> do this # ninety-billion <pause dur="0.4"/> option search every time you make a move <pause dur="4.3"/> and # <pause dur="0.7"/> de Groot's colleagues found further evidence to support this in an experiment that involved chess master and expert players' <pause dur="0.4"/> recall of board positions <pause dur="2.1"/> they found that if the position was from an actual game <pause dur="0.7"/> than the grand masters

<pause dur="0.4"/> were <pause dur="0.3"/> significantly <pause dur="0.4"/> better <pause dur="0.8"/> at remembering <pause dur="0.3"/> the position <pause dur="4.1"/> but <pause dur="0.7"/> if <pause dur="0.2"/> the position was random <pause dur="2.5"/> so you just <pause dur="0.2"/> pick your chess pieces <pause dur="0.2"/> whack them anywhere on the board <pause dur="0.2"/> nothing to do with the game <pause dur="0.7"/> then you found no difference <pause dur="0.3"/> between <pause dur="0.6"/> # the two <pause dur="0.2"/> <trunc>exp</trunc> <pause dur="0.5"/> expert groups no difference between the experts and the grand masters <pause dur="1.4"/> okay so <trunc>y</trunc> <pause dur="0.2"/> grand masters only get their advantage <pause dur="0.3"/> if they're trying to memorize an actual board position <pause dur="0.2"/> and they argued <pause dur="0.3"/> that <trunc>tha</trunc> this is because <pause dur="0.2"/> the grand masters simply <pause dur="0.3"/> knew more <pause dur="0.7"/> game <pause dur="0.4"/> positions <pause dur="6.7"/> so <pause dur="2.0"/> according to <pause dur="0.3"/> move on to Chase and Simon who followed up this work what <pause dur="0.2"/> differentiates <pause dur="0.7"/> expert chess players from novice chess players <pause dur="0.8"/> is that experts have stored <pause dur="0.4"/> and

organized in memory <pause dur="0.5"/> many tens of thousands <pause dur="0.5"/> of <pause dur="0.2"/> different <pause dur="0.2"/> game <pause dur="0.2"/> positions <pause dur="2.4"/> and the idea is when they saw sensible game positions they could use the knowledge they had <pause dur="0.3"/> in memory <pause dur="0.9"/> to help them remember <pause dur="0.2"/> the various <pause dur="0.3"/> positions <pause dur="0.2"/> as integrated <pause dur="0.3"/> organized <pause dur="0.4"/> chunks <pause dur="0.7"/> of <pause dur="0.9"/><kinesic desc="reveals covered part of transparency" iterated="n"/> information <pause dur="3.7"/> and <pause dur="0.3"/> this is the sort of thing that Chase and Simon <pause dur="0.2"/> nineteen-seventy-three <pause dur="0.6"/> were looking at <pause dur="1.6"/> okay if you remember back to # <pause dur="0.3"/> about this time last year <pause dur="1.0"/> # <pause dur="0.7"/> and my first year practical on <pause dur="0.2"/> memory chunking <pause dur="1.1"/> you'll remember that you could remember <pause dur="0.4"/> the same <pause dur="0.3"/> stimuli <pause dur="0.4"/> better <pause dur="0.6"/> when they were <pause dur="0.2"/> sort of chunked into groups <pause dur="0.3"/> that linked in <pause dur="0.7"/> with your own personal knowledge basis so remember i presented you with a list of <pause dur="0.5"/> sort of letters <pause dur="0.4"/> when i arranged them in one group in in certain groups they made no sense and so you're trying to remember each individual letter <pause dur="0.4"/> when i arranged them in other groups <pause dur="0.2"/> they actually formed <pause dur="0.2"/> # acronyms <pause dur="0.3"/> <unclear>sort of</unclear> common acronyms # <pause dur="0.4"/> so that instead of having to remember every single letter <pause dur="0.2"/> you could just remember <pause dur="0.4"/>

the fewer number <pause dur="0.2"/> of acronyms <pause dur="1.4"/> okay so this is all Miller's stuff on there being room for <pause dur="0.5"/> seven plus-or-minus two <pause dur="0.3"/> items in short term memory <pause dur="1.4"/> but the size <pause dur="0.4"/> of <pause dur="0.4"/> what <pause dur="0.2"/> one of those items is <pause dur="0.3"/> depends on your <pause dur="0.4"/> personal <trunc>knowle</trunc> knowledge <pause dur="1.3"/> and Chase <trunc>an</trunc> <trunc>ar</trunc> Simon argued that <pause dur="0.6"/> chess players are chunking <pause dur="0.5"/> the board <pause dur="0.3"/> and the more expert you get at chess <pause dur="0.3"/> the bigger more complex <pause dur="0.3"/> your chunks <pause dur="0.6"/> become <pause dur="6.0"/> and # <pause dur="0.4"/> chess experts were found <pause dur="1.0"/> to be able to memorize board positions <pause dur="1.6"/> by breaking them down into seven or so <pause dur="0.2"/> familiar patterns or units <pause dur="1.6"/> and the difference between the <pause dur="0.2"/> novice and expert <pause dur="0.4"/> lay in the amount of information they could hold <pause dur="0.2"/> in one chunk <pause dur="15.2"/> so in # one experiment they did <pause dur="0.9"/> people had two chess boards in front of them <pause dur="0.8"/> on the first board was an arrangement of pieces <pause dur="0.9"/> and the second board <pause dur="0.3"/> was blank <pause dur="2.1"/> and what participants had to do was recreate the position they saw on the first board <pause dur="1.5"/> with the <trunc>pos</trunc> # <pause dur="0.5"/> sorry <pause dur="0.2"/> say that again <pause dur="0.4"/> what participants had to do was recreate the position they could see on

the first board <pause dur="0.3"/> on the second board <pause dur="1.2"/> and what they measured <pause dur="0.4"/> was the number of glances <pause dur="0.3"/> that these people <pause dur="0.5"/> had between the two boards <pause dur="0.8"/> and how much <pause dur="0.6"/> they remembered <pause dur="0.2"/> in <pause dur="0.2"/> each of these glances <pause dur="1.9"/> okay <pause dur="0.5"/> so <pause dur="0.2"/> they argued that each glance between these two boards could be viewed as <pause dur="0.2"/> a chunk <pause dur="3.9"/> and better players could <trunc>rem</trunc> <trunc>r</trunc> recognize a chunk faster <pause dur="1.5"/> and <pause dur="0.2"/> their chunks were also <pause dur="0.2"/> bigger <pause dur="10.8"/> and # work <pause dur="0.2"/> published <pause dur="1.0"/> in nineteen-ninety-six by <sic corr="Gobet">Gobel</sic><pause dur="0.2"/> and Simon <pause dur="5.2"/><kinesic desc="changes transparency" iterated="y" dur="10"/> confirms this <pause dur="9.7"/> and # <pause dur="0.7"/> they demonstrated that <pause dur="0.2"/> retrieval processes involving <pause dur="0.8"/> recognition of board <pause dur="0.2"/> arrangements <pause dur="1.8"/> are actually instrumental <pause dur="1.9"/> in grand master level chess players' success <pause dur="0.4"/> compared with <pause dur="0.7"/> # <pause dur="0.9"/> less experienced <pause dur="0.4"/> other players <pause dur="7.8"/> okay so even when <pause dur="0.2"/> the grand masters were time constrained <pause dur="0.6"/> so they couldn't engage their looking ahead processes their performance isn't much affected <pause dur="0.9"/> okay <pause dur="0.6"/> and they argued <pause dur="0.7"/> this time that this suggests that an organized <pause dur="0.2"/> knowledge system <pause dur="1.0"/> is relatively more important to <pause dur="0.2"/> experts' performance <pause dur="0.5"/> than even the processes

involved in predicting future moves <pause dur="0.4"/> okay so <pause dur="0.3"/> the key thing is organization <pause dur="0.4"/> of information <pause dur="0.7"/> # <pause dur="0.3"/> organization of knowledge <pause dur="6.8"/> having said that <pause dur="0.6"/> # Holding and Reynolds <pause dur="2.2"/> did experiment in nineteen-<pause dur="0.2"/>eighty-two <pause dur="0.7"/> # they reran de Groot's experiment using a random board position <pause dur="0.5"/> okay so again chess board <pause dur="0.4"/> and you just whack on the pieces at random <pause dur="0.4"/> nothing to do with the game <pause dur="1.2"/> and <pause dur="0.4"/> by comparing chess players of different levels <pause dur="0.7"/> on the move they would make from <pause dur="0.5"/> the random <pause dur="0.2"/> position <pause dur="1.7"/> they found that the experts produced better moves <pause dur="0.4"/> on <pause dur="0.3"/> even from these totally random board positions <pause dur="0.8"/> okay <pause dur="0.3"/> so the idea is here is that # <pause dur="0.4"/> the experts couldn't possibly have the board position previously analysed <pause dur="0.3"/> because it was a totally random board position <pause dur="1.1"/> # but they <trunc>inst</trunc> still generated better moves on from those random board positions so <pause dur="0.4"/> they argued that # <pause dur="1.3"/><kinesic desc="reveals covered part of transparency" iterated="n"/> we shouldn't be so one-sided about this <pause dur="0.3"/> and <pause dur="2.4"/> there is a important element of chess expertise that does involve <pause dur="0.8"/> evaluating <pause dur="0.2"/> board position as well

as <pause dur="0.5"/> remembering it <pause dur="4.1"/> okay <pause dur="1.2"/> right so <pause dur="0.4"/> half an hour in so # <pause dur="0.6"/> kind of thought <pause dur="0.4"/> # <trunc>s</trunc> skill <pause dur="0.6"/> and learning things this is when your attention should be sort of plummeting downhill <pause dur="0.3"/> so i shall attempt to raise it again by doing a quick demonstration <pause dur="0.2"/> of <pause dur="0.4"/> one of these experiments <pause dur="0.5"/> if i get you to stick your hand up if you've ever been in a chess club <pause dur="2.1"/> <vocal desc="laughter" n="sl" iterated="y" dur="1"/> can i get <trunc>y</trunc> stick your hand up if you've ever played chess before <pause dur="0.8"/><vocal desc="laughter" n="sl" iterated="y" dur="1"/> bollocks <vocal desc="laughter" n="sl" iterated="y" dur="2"/> <shift feature="voice" new="laugh"/> right<shift feature="voice" new="normal"/><pause dur="0.2"/> okay could you put your hand up if you think you're intermediate chess player <pause dur="0.2"/> <trunc>y</trunc> <pause dur="1.3"/> sort of okayish <pause dur="1.8"/> it's getting tricky i'm going to have to pick at random <unclear>i think</unclear><pause dur="0.2"/> okay put your hand up if you've never ever played chess before <pause dur="1.3"/><kinesic desc="put hands up" n="ss" iterated="n"/> right out of those people who couldn't even name the pieces on a chess board <pause dur="1.3"/><kinesic desc="put hands up" n="ss" iterated="n"/> brilliant <pause dur="0.4"/> # <trunc>cou</trunc> could i can i get you to come out <event desc="walks to front of class" n="sf1209" iterated="y" dur="3"/> <pause dur="2.1"/> right <pause dur="0.5"/> okay and out of the rest of you i need to find someone who # <pause dur="0.4"/> okay who who's been playing since they were # a child who has played as chess as a child and has played recently as

well <pause dur="0.5"/> can you stick your hand up <pause dur="2.0"/><kinesic desc="put hands up" n="ss" iterated="n"/> <vocal desc="laughter" n="sl" iterated="y" dur="1"/> who's played chess within the last four years <kinesic desc="put hands up" n="ss" iterated="n"/> <vocal desc="laughter" n="sl" iterated="y" dur="3"/> ah brilliant <gap reason="name" extent="1 word"/></u><u who="sf1210" trans="latching"> oh no</u> <pause dur="0.2"/> <u who="nm1208" trans="pause"> it's all right <vocal desc="laughter" n="sl" iterated="y" dur="1"/> this is undermining <pause dur="0.3"/> okay what i can i <kinesic desc="turns on overhead projector showing transparency"/> get you to stand over there and look in that direction <pause dur="0.6"/> okay <pause dur="0.9"/> right </u> <u who="sf1210" trans="latching"> look in which direction </u> <u who="nm1208" trans="overlap"> <trunc>s</trunc> <trunc>o</trunc> over there <pause dur="0.2"/> just just don't don't look over here <pause dur="0.2"/> okay <vocal desc="laughter" n="sf1210" iterated="y" dur="1"/> <pause dur="0.5"/> and what i'm going to show you <pause dur="0.5"/> i've # what i've designed here is # my acme patent overhead <pause dur="0.4"/> chess set <pause dur="2.7"/><vocal desc="laughter" n="ss" iterated="y" dur="3"/> carefully constructed there <pause dur="1.1"/><kinesic desc="changes transparency" iterated="y" dur="2"/> okay <pause dur="0.6"/> and basically what i'm going to do is # i've got a chess board position <pause dur="0.3"/> from a famous game <pause dur="0.4"/> for <gap reason="name" extent="2 words"/> versus <gap reason="name" extent="1 word"/> played this Tuesday <vocal desc="laughter" n="ss" iterated="y" dur="2"/> <pause dur="0.7"/> and # if you come and stand here <pause dur="1.7"/><event desc="walks to position" n="sf1209" iterated="y" dur="2"/> # i'm going to show that for thirty seconds </u> <pause dur="0.3"/> <u who="nm1208" trans="pause"> oh God </u> <pause dur="0.3"/> <u who="nm1208" trans="pause"> and then i'm going to cover it up <pause dur="0.2"/> and then i want you to recreate that position on that board <pause dur="0.4"/> as <trunc>f</trunc> <pause dur="0.2"/> as # fast as you can <pause dur="0.3"/> and you've got about two minutes or just tell me <trunc>y</trunc> <pause dur="0.2"/> when you give up <pause dur="0.4"/> okay <vocal desc="laughter" n="sl" iterated="y" dur="3"/> right <pause dur="1.4"/> and we'll <trunc>s</trunc> see <pause dur="0.2"/> okay so you remember in the original experiment this was actually a comparison between # expert club

players <pause dur="0.3"/> and a grand master <pause dur="0.2"/> so we're actually looking at very different levels of expertise but we can see whether we can <pause dur="0.5"/> sort of replicate the same sort of effect so are you ready <pause dur="0.7"/><kinesic desc="starts stopwatch" iterated="n"/><kinesic desc="memorizes board position" n="sf1209" iterated="y" dur="30"/> and here goes <pause dur="2.3"/> for those of you can't see this board position this is probably a very <shift feature="voice" new="laugh"/>boring<shift feature="voice" new="normal"/> part of the lecture <vocal desc="laughter" n="sf1211" iterated="y" dur="2"/> <vocal desc="laughter" iterated="y" dur="1"/> <pause dur="1.5"/> okay <pause dur="0.7"/> <trunc>t</trunc> <pause dur="0.2"/> ten seconds <pause dur="9.0"/> twenty seconds <pause dur="8.8"/> thirty seconds okay off you go <pause dur="3.5"/> <kinesic desc="moves chess pieces to memorized positions" n="sf1209" iterated="y" dur="53"/> okay <pause dur="1.7"/> and of course we have the additional problem here is that # these pieces are really fiddly to move around <pause dur="6.5"/><gap desc="inaudible" extent="1 word"/><pause dur="9.2"/> okay and what we're going to try and do here is # score this by just counting the number of <pause dur="0.6"/> pieces <pause dur="0.2"/> that she gets in the right place <pause dur="15.8"/> okay if you wanted to do this scientifically of course then # i'd actually </u><u who="sf1209" trans="overlap"> <gap reason="inaudible" extent="1 sec"/> guessed the rest </u><pause dur="0.9"/> <u who="nm1208" trans="pause"> you're guessing the rest okay that's <pause dur="0.7"/> right <pause dur="0.4"/> # <pause dur="0.6"/> i don't think you did too badly there <pause dur="1.1"/> okay # right looking my original piece you've got all the <pause dur="1.0"/> one two three four five six seven <pause dur="1.2"/> eight <pause dur="0.6"/> nine <pause dur="3.0"/> you get nine points <pause dur="1.3"/> brilliant thanks very much indeed <vocal desc="laughter" iterated="y" dur="1"/> that's it <vocal desc="laughter" n="sf1209" iterated="y" dur="1"/> <kinesic desc="applause" n="sl" iterated="y" dur="4"/> right <pause dur="1.1"/> okay i'll just # quickly scramble this # <pause dur="11.4"/> <kinesic desc="mixes up pieces on transparency" iterated="y" dur="11"/> all right <pause dur="1.6"/>

okay <pause dur="2.3"/> right # and could i have my # chess grand master <vocal desc="laughter" n="sl" iterated="y" dur="3"/> </u><pause dur="1.4"/> <u who="sf1210" trans="pause"> <gap reason="inaudible" extent="2 secs"/></u> <pause dur="0.4"/> <u who="nm1208" trans="pause"> okay <pause dur="0.8"/> right do you want to stand here <pause dur="0.5"/> and <pause dur="0.2"/> first i'm going to show you this # real chess board position <pause dur="0.2"/> for thirty seconds then i want you to <pause dur="0.6"/> # going to cover it up and then you've got about two minutes to reproduce as much as you can remember <pause dur="0.3"/> okay </u> <u who="sf1210" trans="latching"> what on this </u><pause dur="0.4"/> <u who="nm1208" trans="pause"> # yeah well on # yeah that's right yeah <pause dur="0.7"/> okay you're ready <kinesic desc="starts stopwatch" iterated="n"/></u><pause dur="1.0"/> <u who="sf1210" trans="pause"> sorry</u> <pause dur="0.4"/> <u who="nm1208" trans="pause"> <vocal desc="laugh" iterated="n"/> <pause dur="0.6"/> and <pause dur="0.6"/> go <kinesic desc="memorizes board position" n="sf1210" iterated="y" dur="30"/> <pause dur="1.2"/> <gap reason="inaudible" extent="1 word"/><vocal desc="laughter" n="sl" iterated="y" dur="2"/> # # <kinesic desc="starts stopwatch" iterated="n"/> <pause dur="11.9"/> okay that's ten seconds <pause dur="13.7"/> okay twenty-seven seconds three <pause dur="0.7"/> two <pause dur="0.5"/> one <pause dur="0.3"/> go <kinesic desc="stops stopwatch" iterated="n"/><pause dur="0.8"/> okay right get you to reproduce it <pause dur="2.7"/><kinesic desc="moves chess pieces to memorized positions" n="sf1210" iterated="y" dur="1:12"/> okay so according to <pause dur="0.2"/> de Groot someone who has a basic knowledge of chess <pause dur="0.6"/> # <pause dur="0.3"/> should be able to # remember <pause dur="0.4"/> slightly more <pause dur="0.2"/> positions <pause dur="0.3"/> if you noticed with the <pause dur="0.3"/> on the on the actual sort of stimuli that the de Groot was using <pause dur="0.3"/> # <pause dur="0.4"/> <trunc>a</trunc> actually no sorry this is <trunc>ch</trunc> # Chase and Simon this is <pause dur="0.4"/> # <pause dur="0.3"/> obviously because they're using much more higher level of chess players # the actual chess boards were like <pause dur="0.2"/> much further into the game which is <trunc>wh</trunc> <pause dur="0.4"/> i deliberately chose one where <trunc>we</trunc> we're only we're only

about sort of think ten moves into the game there <pause dur="0.6"/> or something like that so a lot of the pieces are in their original positions <pause dur="1.0"/> but # <pause dur="0.9"/> <vocal desc="laughter" iterated="y" dur="1"/> </u> <pause dur="7.2"/> <u who="sf1210" trans="pause"> i've mixed the colours up does it matter </u> <pause dur="0.2"/> <u who="nm1208" trans="pause"> no <pause dur="1.2"/><vocal desc="laughter" iterated="y" dur="2"/> okay in the scientist's <shift feature="voice" new="laugh"/> conditions <shift feature="voice" new="normal"/> then we would care about the colours <vocal desc="laughter" n="sl" iterated="y" dur="3"/> <shift feature="voice" new="laugh"/> and # <shift feature="voice" new="normal"/> <pause dur="0.5"/> okay and # <pause dur="0.6"/> also if this is is this is proper science of course i'll be testing like fifty people in each group <pause dur="0.4"/> and # and also # control for manual dexterity in moving bits of tiny acetate <shift feature="voice" new="laugh"/>around<shift feature="voice" new="normal"/> <vocal desc="laughter" n="sl" iterated="y" dur="2"/> </u><u who="sf1212" trans="latching"> probably better than you did it </u> <pause dur="0.4"/> <u who="nm1208" trans="pause">

okay that's cool thanks very much indeed <pause dur="0.6"/> right so <pause dur="3.1"/> right # we have <pause dur="0.3"/> one <pause dur="4.0"/> two <vocal desc="laugh" iterated="n"/> no <vocal desc="laughter" n="sl" iterated="y" dur="2"/> three four five <pause dur="2.2"/> six <pause dur="0.6"/> seven eight nine ten eleven twelve <pause dur="1.9"/> twelve # <pause dur="0.5"/> twelve <pause dur="0.6"/> ha <vocal desc="laughter" n="ss" iterated="y" dur="1"/> <pause dur="0.9"/> <kinesic desc="applause" n="sl" iterated="y" dur="3"/> hurray <pause dur="1.5"/> <vocal desc="laughter" dur="1" iterated="y"/> i don't know whether that <shift feature="voice" new="laugh"/> means it's <shift feature="voice" new="normal"/> <trunc>stati</trunc> statistically significant thanks thank you very much you two <pause dur="0.8"/> brilliant <pause dur="1.8"/> # <pause dur="2.1"/> okay <pause dur="0.3"/> <shift feature="voice" new="laugh"/>i'm not<shift feature="voice" new="normal"/> but i'm not going to pretend that's science in action there <vocal desc="laughter" n="sl" iterated="y" dur="3"/> <pause dur="1.3"/> okay <pause dur="1.5"/><kinesic desc="changes transparency" iterated="y" dur="2"/> right so there we have chess nicely defined <pause dur="0.3"/> domain <pause dur="0.4"/> but <pause dur="0.4"/> the problem is <pause dur="1.8"/> how do we know <pause dur="0.3"/> that other more complicated <pause dur="0.2"/> and sort of every day <pause dur="0.4"/> domains <pause dur="0.4"/> # actually <pause dur="0.2"/> involving the same sort of processes <pause dur="0.8"/> in the real world problems are seldom as

well defined as they are in chess <pause dur="0.2"/> so what research is there looking at expertise <pause dur="0.4"/> in domains with less rigid rules <pause dur="1.4"/> and in the literature areas commonly study studied <pause dur="0.8"/> involve <pause dur="1.6"/> <kinesic desc="changes transparency" iterated="y" dur="5"/> things like <pause dur="2.1"/> chess <pause dur="0.3"/> medicine <pause dur="0.3"/> and <pause dur="0.6"/> computing <pause dur="3.9"/> so many of the features that characterize the expert chess players <pause dur="0.6"/> also seem to characterize experts in <pause dur="0.4"/> other domains <pause dur="1.2"/> for example the idea of information chunking <pause dur="0.4"/> was found to be important in the domain of computer programming <pause dur="1.3"/> computer programming it's been suggested that expert programmers <pause dur="0.9"/> have <pause dur="0.3"/> large chunks of code <pause dur="0.7"/> in memory that they can rearrange <pause dur="0.7"/> in order to solve a problem <pause dur="1.6"/> so the idea is that expert programmers can remember more code <pause dur="0.3"/> than novices <pause dur="0.5"/> and as Chase and Simon then showed with chess <pause dur="0.5"/> they can <pause dur="0.2"/> fit more information <pause dur="0.2"/> into one <pause dur="0.3"/> sort of memory chunk <pause dur="0.5"/> of code <pause dur="0.2"/> if you like <pause dur="3.4"/> okay <pause dur="0.6"/> as i said # <pause dur="1.9"/> well that was research by Adelson in nineteen-eighty-one <pause dur="2.3"/> one way of # analysing difference between experts and novices <pause dur="0.2"/> as i said previously <pause dur="0.2"/> is to get people to talk about

what they're thinking of <pause dur="0.5"/> when their <trunc>solv</trunc> when they solve a problem <pause dur="0.7"/> and these are verbal protocols <pause dur="1.5"/> and # <pause dur="1.1"/> people have done this looking at these sort of domains <pause dur="0.6"/> and they've compared they've transcribed the commentaries and compared the statements people are making <pause dur="0.6"/> # between the expert and novice groups and they've also compared the length of time <pause dur="0.3"/> people have spent on various aspects <pause dur="0.6"/> of the problems <pause dur="0.9"/> and also the relationship between these sort of strategies <pause dur="0.5"/> used <pause dur="0.3"/> and the solutions reached <pause dur="2.9"/><kinesic desc="changes transparency" iterated="y" dur="5"/> and <pause dur="1.3"/> it is using that sort of methods <pause dur="0.6"/> that # <pause dur="0.5"/> Glaser and Chi <pause dur="1.0"/> nineteen-eighty-eight <pause dur="9.2"/> found that # <pause dur="0.6"/> one difference <pause dur="0.2"/> between experts and novices <pause dur="0.2"/> is their different schemas <pause dur="1.1"/> for <pause dur="0.5"/> # solving problems within their own domain of expertise <pause dur="0.5"/> okay <pause dur="0.5"/> this bit of # overhead is actually on the handout so you don't need to scribble it down <pause dur="0.7"/> okay <pause dur="0.3"/> and so our schema <pause dur="0.5"/> what i mean by a schema <pause dur="0.2"/> is a sort of a plan <pause dur="0.2"/> an outline a structure <pause dur="0.2"/> framework a program <pause dur="0.7"/> okay so <pause dur="0.4"/> in this sort of context <pause dur="0.2"/> think of a schema <pause dur="0.5"/> as a sort

of cognitive mental plan <pause dur="0.4"/> sort of guide for action <pause dur="1.0"/> okay especially in this case some sort of organized <pause dur="0.2"/> framework <pause dur="0.8"/> for solving problems <pause dur="1.1"/> and the <pause dur="0.2"/> schemas <pause dur="0.3"/> of experts <pause dur="1.0"/> have been argued to involve <pause dur="0.2"/> large <pause dur="0.3"/> highly <trunc>intercon</trunc> interrelated <pause dur="0.6"/> units of knowledge <pause dur="3.7"/> which are organized according to <pause dur="0.2"/> underlying structural similarities <pause dur="0.8"/> amongst these knowledge units <pause dur="2.0"/> okay <pause dur="1.8"/> so our experts large <pause dur="0.2"/> highly interconnected units of knowledge <pause dur="0.8"/> and connected together <pause dur="0.4"/> # <pause dur="0.5"/> by underlying structural similarities <pause dur="0.4"/> a contrast to that is the novices <pause dur="0.3"/> # the idea is that the sort of schemas <pause dur="0.2"/> that novices use <pause dur="0.4"/> in a domain <pause dur="0.5"/> are relatively small <pause dur="0.3"/> disconnected bits of information <pause dur="0.7"/> which are organized according to superficial <pause dur="0.5"/> similarities <pause dur="0.2"/> not structural <pause dur="0.4"/> similarities <pause dur="2.7"/> and # <pause dur="0.9"/> Glaser and Chi <pause dur="0.2"/> this is # <pause dur="0.5"/> <trunc>fun</trunc> # <trunc>th</trunc> this bit here is actually from # Steinberg <pause dur="0.3"/> <trunc>nineteen-ninety-ni</trunc> <trunc>wo</trunc> nine which i'll put in the # reprint collection <pause dur="0.7"/> # <pause dur="0.2"/> this is the difference between novices and experts <pause dur="0.2"/> and schemas <pause dur="0.4"/> it can be noted in how they

classify different various problems <pause dur="1.0"/> # <pause dur="0.9"/> the idea's that experts and novices also describe the essential nature of various problems differently <pause dur="1.1"/> and <pause dur="0.3"/> determine how to solve <pause dur="0.5"/> various problems <pause dur="0.6"/> differently <pause dur="0.9"/> okay so <pause dur="0.3"/> three key expert <pause dur="0.2"/> novice differences <pause dur="5.8"/> and another thing that # researchers have found <pause dur="0.5"/> is that experts tend to spend much more time <pause dur="0.2"/> determining how <pause dur="0.5"/> to <pause dur="0.3"/> represent a problem <pause dur="0.2"/> than novices <pause dur="0.8"/> that's research by <pause dur="0.5"/><kinesic desc="reveals covered part of transparency" iterated="n"/> Lesgold et al <pause dur="0.4"/> nineteen-eighty-eight <pause dur="3.0"/> okay so experts spend more time representing the problem <pause dur="0.3"/> but <pause dur="0.2"/> they spent less time <pause dur="0.3"/> implementing <pause dur="0.4"/> that <trunc>solu</trunc> <pause dur="0.2"/> the solution <pause dur="0.4"/> or the strategy for <trunc>th</trunc> the solution <pause dur="7.8"/> and in contrast <pause dur="0.2"/> novices tend to dive into a problem <pause dur="0.3"/> without so much of this <pause dur="0.5"/> # <pause dur="0.3"/> initial analysis and hence spent much more time <pause dur="0.3"/> trying to figure out <pause dur="0.4"/> solution <pause dur="0.2"/> to the problem <pause dur="3.4"/> so differences between experts and novices in these sort of domains <pause dur="0.3"/> in their expenditure of time could be viewed in terms of the focus <pause dur="0.5"/> and the direction of their problem solving <pause dur="0.3"/> experts

spend more time figuring out what they <pause dur="0.3"/> already know about the problem <pause dur="0.4"/> and how the information given in a problem <pause dur="0.2"/> maps on to what <pause dur="0.7"/> they already know <pause dur="0.9"/> and the idea is once the expert finds a previously existing strategy for solving <pause dur="0.4"/> the problem they can just put it down out of memory <pause dur="0.4"/> and implement it <pause dur="0.6"/> without too much bother <pause dur="1.9"/> okay <pause dur="0.9"/> and <pause dur="0.2"/> another way to describe that <pause dur="0.2"/> is that experts are <pause dur="0.3"/> working forward <pause dur="1.4"/> from the given information <pause dur="0.4"/> to find <pause dur="0.7"/> their unknown information <pause dur="0.6"/> okay <pause dur="5.4"/><kinesic desc="changes transparency" iterated="y" dur="2"/> okay so they're going from what do i know to what do i need to find out <pause dur="1.4"/> implementing the correct sequence of steps based on strategies they've retrieved <pause dur="0.2"/> from their <pause dur="0.2"/> schemas in long term memory <pause dur="0.9"/> okay in contrast <pause dur="1.1"/> to this <pause dur="0.3"/> the novices <pause dur="0.5"/> tend to work <pause dur="0.2"/> backwards <pause dur="0.7"/> so what they tend to do is generate a set of alternative <pause dur="0.4"/> solutions <pause dur="0.9"/> to start with and then trying to work out <pause dur="0.4"/> which one would be the best one to proceed with <pause dur="0.7"/> okay so little time is spent trying to represent the problem initially <pause dur="1.0"/> and # <pause dur="0.4"/> to illustrate this difference between forward

processing and <pause dur="0.2"/> backwards processing here is an example <pause dur="0.2"/> that's in Steinberg <pause dur="2.3"/> imagine that we have an expert doctor <pause dur="0.5"/> and <pause dur="0.3"/> a totally green <pause dur="0.5"/> # novice <pause dur="0.3"/> <trunc>n</trunc> medical student <pause dur="0.6"/> and they're both presented with a patient <pause dur="0.5"/> who has a series of symptoms <pause dur="0.9"/> first of all what does the novice do <pause dur="0.4"/> well he's not sure <pause dur="0.4"/> exactly what to make of the symptoms and so what he does is goes and orders a <pause dur="0.3"/> whole long series of expensive <pause dur="0.4"/> # <pause dur="0.2"/> medical tests to be done <pause dur="0.5"/> in the hope <pause dur="0.2"/> that once he's got <pause dur="0.2"/> the full <pause dur="0.3"/> information in front of him full symptomatic information he may be able to <pause dur="0.6"/> then <pause dur="0.2"/> go ahead and make his diagnosis <pause dur="0.5"/> and he realizes that this illness could be <pause dur="0.2"/> any number of a wide range of things <pause dur="0.3"/> and he works backwards <pause dur="0.4"/> to try and <pause dur="0.3"/> sort eliminate them and work out exactly which one <pause dur="0.2"/> it is <pause dur="1.6"/> our more experienced doctor on the other hand <pause dur="0.3"/> is far more likely to recognize <pause dur="0.3"/> the initial <pause dur="0.6"/> set of symptoms he's being presented with <pause dur="0.8"/> as being as fitting <pause dur="0.2"/> a diagnostic pattern <pause dur="0.3"/> or one of a small number of patterns that she holds in her <pause dur="0.5"/>

long term memory <pause dur="2.1"/> she's therefore worked forward from these initial symptoms <pause dur="0.2"/> to a much smaller <pause dur="0.6"/> set of possible <pause dur="0.2"/> illnesses <pause dur="0.3"/> and therefore she only needs <pause dur="0.4"/> a very small number <pause dur="0.3"/> of highly targeted tests <pause dur="0.5"/> to choose <pause dur="0.3"/> the correct <pause dur="0.2"/> diagnosis <pause dur="0.2"/> from among the <pause dur="0.2"/> limited number of possibilities that she's generated <pause dur="3.8"/> okay <pause dur="0.6"/> # <pause dur="0.5"/> okay a word of warning with that though <pause dur="0.3"/> is that obviously that's a that's a simple way <pause dur="0.3"/> of describing it and as you can imagine in real life <pause dur="0.3"/> get's much more <pause dur="0.4"/> fiddly than that <pause dur="0.9"/> okay those are sort of very sort of gross <pause dur="1.4"/> it's a very gross generalization on how <pause dur="0.9"/> people operate 'cause obviously as i described at the beginning it depends on how you define your expert <pause dur="0.4"/> and how you <pause dur="0.4"/> define your novice <pause dur="3.6"/> okay <pause dur="0.4"/> and # <pause dur="1.9"/> another thing to bear in mind <pause dur="0.2"/> is <pause dur="0.6"/> what a <trunc>w</trunc> first of all well <trunc>wh</trunc> <trunc>d</trunc> <trunc>w</trunc> how do we define our expertise and also <pause dur="0.2"/> how <pause dur="0.5"/> people in the study have chosen to measure <pause dur="0.7"/> that expertise <pause dur="2.1"/> and <pause dur="0.7"/> this actually makes the whole area of expertise quite tricky <pause dur="0.5"/> and when you're actually

looking at research in this area <pause dur="0.3"/> you should bear <pause dur="0.2"/> that sort of thing in mind <pause dur="0.3"/> okay <pause dur="0.3"/> because despite the example i've just given you <pause dur="0.7"/> # what you actually find with doctors <pause dur="0.4"/> is that <pause dur="2.6"/><kinesic desc="changes transparency" iterated="y" dur="9"/>for most <pause dur="0.2"/> cases <pause dur="3.3"/> you don't actually get that much difference between novice and expert doctors <pause dur="1.2"/> so it's been found that <pause dur="0.3"/> diagnostic performance doesn't actually seem to improve <pause dur="0.5"/> much beyond the first year <pause dur="0.4"/> of residency <pause dur="0.4"/> for <pause dur="0.2"/> typical <pause dur="0.2"/> diagnostic cases <pause dur="3.4"/> okay and # <pause dur="0.2"/> there are some small differences between doctors with different lengths of experience for everyday <trunc>agn</trunc> diagnosis <pause dur="0.8"/> # <pause dur="0.7"/> though it has been shown <pause dur="0.3"/> there are tend to be much bigger differences <pause dur="0.3"/> as soon as you start moving on to more difficult cases <pause dur="0.9"/> okay <pause dur="11.3"/><kinesic desc="reveals covered part of transparency" iterated="n"/> and <pause dur="0.7"/> Boshuizen <pause dur="0.3"/> and Schmidt nineteen-ninety-two <pause dur="2.9"/> found that the expert doctors seem to have <pause dur="0.3"/> easier access to higher level <pause dur="0.8"/> structured diagnostic information <pause dur="0.6"/> whereas the medical students <pause dur="0.6"/> tended to have to go through this sort of cumbersome biomedical reasoning <pause dur="0.5"/> to get to their solutions <pause dur="0.6"/> okay <pause dur="1.3"/> but <pause dur="0.2"/> say we've got to be

very careful that we don't <pause dur="0.2"/> overgeneralize here <pause dur="0.2"/> because it matters what we're actually calling a novice doctor here and what we're calling an expert doctor <pause dur="0.4"/> and <pause dur="0.2"/> how we're measuring <pause dur="0.6"/> their expertise <pause dur="2.1"/> another example <pause dur="0.6"/> of a <pause dur="0.9"/><kinesic desc="changes transparency" iterated="y" dur="10"/> skilled domain where the problem may not be <pause dur="0.2"/> as well defined as we would like <pause dur="0.6"/> is <pause dur="0.2"/> physics <pause dur="3.2"/> and # <pause dur="0.9"/> it's covered by a lot of research by Chi et al <pause dur="1.7"/>and they tested # expert and novice physicists on a range of physics problems <pause dur="1.0"/>

and they found that the novices tend to use the surface features <pause dur="0.3"/> of the problem <pause dur="0.5"/> while the experts <pause dur="0.4"/> # encode far more deeply <pause dur="2.5"/> # in this domain experts solved problems four times faster <pause dur="0.2"/> than novices <pause dur="0.6"/> though they spent longer analysing the problems <pause dur="0.3"/> and Chi et al argued that these differences <pause dur="0.4"/> between novices and experts were <pause dur="0.3"/> as a result <pause dur="1.1"/> of differences in strategy <pause dur="0.5"/> and <pause dur="0.2"/> # knowledge <pause dur="2.1"/> so <pause dur="0.8"/> like with the chess <pause dur="0.4"/> the experts who use their superior knowledge of previous problems <pause dur="0.8"/> # <pause dur="0.2"/> but # is found that the experts not only had a greater quantity of knowledge <pause dur="0.2"/>

they also organized <pause dur="0.3"/> that knowledge <pause dur="0.3"/> better <pause dur="1.7"/> and # <pause dur="0.3"/> experts also tend to exhibit superior memory <pause dur="0.2"/> on unexpected recall tasks <pause dur="0.3"/> possibly as a result of their deeper <pause dur="0.4"/> encoding <pause dur="0.5"/> of the problem <pause dur="3.5"/> okay <pause dur="0.3"/> and <pause dur="2.1"/><kinesic desc="reveals covered part of transparency" iterated="n"/> more research by Larkin et al <pause dur="0.2"/> where they <pause dur="0.2"/> took people <pause dur="0.5"/> with equivalent knowledge in solving <pause dur="0.2"/> physics problems <pause dur="0.2"/> but who differed on their level differed on their level of expertise and they found that <pause dur="0.3"/> when we actually controlled for the amount of knowledge people knew <pause dur="0.5"/> the # the experts still showed a superior performance <pause dur="0.6"/> okay <pause dur="0.3"/> so <pause dur="0.3"/> it's not <trunc>th</trunc> the <trunc>am</trunc> not just the amount of knowledge <pause dur="0.2"/> it's how you structure that knowledge <pause dur="0.2"/> which is important <pause dur="0.3"/> and then <pause dur="0.2"/> they described it as <pause dur="0.2"/> pattern based <pause dur="0.6"/> retrieval <pause dur="0.4"/> from memory <pause dur="0.9"/> and like medicine <pause dur="0.5"/> # they found that <pause dur="0.3"/> experts tended to use a forward working strategy <pause dur="0.6"/> whereas <pause dur="0.2"/> the novices tended to use a backwards <pause dur="0.3"/> working strategy <pause dur="2.6"/> and also this # expertise was found to be very domain <trunc>suspic</trunc> specific <pause dur="0.3"/> so <pause dur="0.2"/> being an expert physicist <pause dur="0.2"/> doesn't make you <pause dur="0.2"/> good at anything

else <pause dur="0.2"/> whatsoever <pause dur="1.1"/> okay and this may reflect the finding <pause dur="0.4"/> that i talked about earlier where basic <pause dur="0.2"/> differences like things like I-Q <pause dur="0.5"/> # have surprisingly little <pause dur="0.2"/> bearing <pause dur="0.3"/> on expert <pause dur="0.3"/> performance <pause dur="4.5"/> okay <pause dur="1.5"/> <kinesic desc="changes transparency" iterated="y" dur="15"/> # running out of time so <pause dur="0.2"/> i'm going to <pause dur="2.6"/> pop on to <pause dur="0.4"/> some people who've tried to summarize <pause dur="0.2"/> this sort of research <pause dur="4.2"/> okay <pause dur="0.3"/> # summarizing this sort of research <pause dur="1.8"/> <gap reason="inaudible" extent="1 word"/><pause dur="0.7"/> is a problem <pause dur="0.2"/> 'cause as i said <pause dur="0.2"/> before we have to be very careful that <pause dur="0.5"/> first of all we know what definition of expert and novice we're using <pause dur="1.1"/> secondly <pause dur="0.3"/> we've got to <pause dur="0.4"/> be very careful about exactly how we're measuring <pause dur="0.2"/> expertise <pause dur="0.5"/> and the third thing we've got to watch out <pause dur="0.6"/> is <pause dur="0.2"/> that we don't overgeneralize from our results <pause dur="0.4"/> okay so it could be <pause dur="0.3"/> that # we say ah this is a <trunc>novi</trunc> this is an expert and the expert can do this this and this <pause dur="0.3"/> but then we discover <pause dur="0.2"/> that's only true of experts <pause dur="0.4"/> # solving maths problems <pause dur="0.2"/> and it has no bearing on experts <pause dur="0.7"/> # doing chess and things like that <pause dur="0.6"/> so you've got to bear all of these sort of these those three things in mind when you're analysing this research <pause dur="0.8"/> but # <pause dur="0.2"/> Glaser and Chi <pause dur="1.8"/> again have come <pause dur="0.2"/> come up with # a list <pause dur="0.8"/> of things they

argued were common <pause dur="0.2"/> between different domains <pause dur="0.2"/> of expertise <pause dur="1.4"/> and <pause dur="0.3"/> these are all things that we've met previously in the lecture <pause dur="1.0"/><kinesic desc="changes transparency" iterated="y" dur="2"/> okay <pause dur="0.2"/> so <pause dur="0.9"/> they argued that # <reading>experts excel mainly in their own domains <pause dur="1.8"/> experts perceive large meaningful patterns in domains <pause dur="0.7"/> experts tend to be faster at solving problems <pause dur="1.1"/> experts have superior short and long term memory <pause dur="1.2"/> experts see and represent problems in their domain at a deeper level than novices <pause dur="0.8"/> and experts spend more time <pause dur="0.4"/> analysing <pause dur="0.4"/> problems <pause dur="0.4"/> qualitatively</reading> <pause dur="1.4"/> and something we haven't talked about is <reading>experts have very strong <pause dur="0.2"/> self-monitoring <pause dur="0.5"/> skills</reading> <pause dur="0.4"/> okay <pause dur="0.5"/> and you'll see that <pause dur="0.5"/> similar pattern reflected <pause dur="0.2"/> in another example i've put in your handout <pause dur="0.5"/> # by Green and Gilhooly who tried to do the same thing and came up with <pause dur="0.3"/> five maxims <pause dur="1.3"/> okay <pause dur="1.2"/> right so we've looked at the differences between novices and experts <pause dur="0.3"/> one point we haven't addressed <pause dur="0.2"/> is how people <pause dur="0.2"/> become <pause dur="0.2"/> experts in the first place so how do you become skilful at something <pause dur="0.5"/> how can <trunc>a</trunc> someone anyone become an expert at something or does <pause dur="0.5"/> # can anyone become an expert at something or does genetics play an important role <pause dur="0.6"/> and those <pause dur="0.2"/> are the issues <pause dur="0.2"/> we'll be looking at <pause dur="0.3"/> next week <pause dur="0.2"/> i'll see you then