Alan Baddeley

Baddeley on management styles

At the end of the meeting on the legacy of the APU, you encouraged people to add further material that might possibly be of interest. One issue that cropped up several times, was the difference in management style between Donald’s directorship and my own, and I thought it might be worth giving my thoughts on this.

As mentioned in the discussion, one crucial difference was the MRC rules for promotion. These were ill defined during Donald’s time, but were assumed to depend upon the Director. The lack of obvious criteria made people worried.

By the time I became Director, the system had changed to one in which one could offer either a three year limited term appointment, or for someone with some post doctoral experience, five years. At the end of this, people could apply for tenure, and were interviewed at Head Office by an independent academic panel, on which the Director sat, but whose decision he could certainly not dictate. People were not in direct competition with each other, and this led to the spirit of cooperation, which as you will recall involved practice interviews, where one’s friends and colleagues proved much tougher interrogators then the proper interview. During my time at the unit, everyone got through. I don’t know what the baseline rate was, but it was certainly not regarded as a foregone conclusion.

I was fortunate in that Donald left me no fewer than seven places to fill at this tenure track level. I had been keeping an eye open for potential appointments in connection with my previous jobs at Sussex and Stirling, and five of the seven were people I already knew, and who knew me (Tony Marcel, Arnold Wilkins, Graham Hitch, Roy and Karalyn Patterson). I also already knew most of the people at the unit, which helped.

I was very fortunate in having spent several years at Sussex, at a time when Stuart Sutherland was building up a very good department. Although a formidable character, and latterly something of a loose cannon, at the time Stuart was a really good head of department, involving his staff in decisions, encouraging everyone to take an active interest in each others research and in seminars, and was very good at encouraging and developing staff. He helped me get a consultancy post with MOD on missile control, which convinced me that you didn’t need to be an expert to be helpful. He encouraged and helped me get research grants, and encouraged my neuropsychological work with Elizabeth Warrington.

I also experienced two contrasting new university departments, UCSD in California where I spent a sabbatical year, and which was full of research stars, and Stirling which was a teaching department where I was employed to build up its research strength. So by the time I got to the APU as Director, I had had a reasonably broad experience, although I had not run anything, apart from a year deputising for Stuart while he was a sabbatical. I remember Alan Kennedy in Dundee asking me was I not terrified to be trying to follow in Donald’s footsteps. My feeling was that if it didn’t work out, I would have no difficulty moving back into an academic job.

I followed Stuart’s example of trying to involve as many people as possible in discussions about decisions, using the Research Coordination Group (RCG) for this purpose. It would have been difficult if the RCG had leaked our discussions on individuals and appointments, but during my time as Director, it appeared to be remarkably leak proof. Of course, that meant that people occasionally got paranoid about the RCG, but at least that comprised people they themselves had elected. In general I tried to play the directorial hand with the cards face up on the grounds that the scenario that people invent through gossip is almost always much more negative then the actual position. In general it seemed to work, which probably reflects the fact the people looking after us in Head Office were consistently supportive, provided you could make a good scientific case, which of course depended on the quality of the people at the Unit.

Of course, all this happened many years ago, and may be largely confabulation, but that’s how I remember it.

Baddeley on Appointments:

Another issue that cropped up in discussion was that of appointment Ann Cutler asked if we had specifically targeted the Sussex psycholinguistic group. The Unit strategy on appointments varied somewhat over the 20 years of my Directorship. The initial appointments were based on my proposals to look at important educational cognitive skills, with Tony Marcel appointed to look at reading, Alan Wing to look at writing, Graham Hitch to look at arithmetic and Arnold Wilkins to look at semantic memory, an area in which he did his PhD, but subsequently largely abandoned. Basically, however people were appointed because they were good cognitive psychologists who I thought would adapt to a regime of combining basic and applied research, so it did not matter too much when we subsequently had to change direction because the board refused to back an educational programme.

In general, we tried to appoint good scientists who would have enough in common with existing staff to integrate, but who were different enough to add something new. A good example of the latter strategy was offered by the case of John Driver who applied both to us and the Cambridge Department. Although clearly extremely good, we decided not to appoint him simply because he was too similar to John Duncan in interests and overall stage of development.

An exception to this general strategy came when we were trying to develop a new area. This was often difficult, because the areas were chosen partly because they were not already well developed. We had great difficulty for example in finding a good research orientated clinical psychologist who might be interested in cognition and emotion. Eventually Fraser Watts contacted us. At the time he did not have a particularly strong research publication record and was principally directing a clinical department at the Institute of Psychiatry. We decided to take a gamble on him, and it worked. Mark Williams chose to give up a tenured lectureship in Newcastle for an untenured post at the APU because it was what he wanted to do. John Teasdale had been part of the group in Oxford on a series of programme grants to Michael Gelder, where the research fellows had reached a stage where Council felt they should become independent. John was not keen on a move, and reluctantly agreed to visit the unit, and to my pleasure and surprise changed his mind, apparently because we were prepared to give him a pretty free hand. Andrew Mathews was another member of the group. I had in fact approached him since I knew him from an earlier joint interest in visual imagery. He had decided to go to the US, but indicated he might be interested later, which proved to be the case.

As you will recall we were not always so successful. I wanted to set up a group interested in decision making, but we were not impressed by any of the relatively small number of people working in this area in the UK, and invited Baruch Fischoff from Oregon to spend a year with us. He fitted in, but decided to return to the US.

With younger staff, we tended to go for the scientifically strongest candidate, provided we felt that they would be able to fit in and broaden and strengthen what we were doing. This meant that we sometimes ended up with someone who was not in the area that we had advertised, an example being Geoff Hinton, who was clearly outstanding, and who assured me that he would not be unduly limited by our lack of computing power. I assumed that he would be able to access any necessary power through the university, but the things he was working on at the time he assured me could be done on an Apple. As you know, it didn’t turn out that way, but it was good to have him.

Over the years we made quite a number of senior appointments. In almost all cases they approached the unit, usually informally, but occasionally in response to a job advert. That happened in the case of Phil Johnson-Laird, and subsequently Ann Cutler and Dennis Norris. Hence my answer to Ann, that we didn’t set out to create a psycholinguistics group, but were very happy to respond very positively to the opportunity.

During the initial years, the balance of positions at particular levels of seniority was relatively clearly specified. Latterly this constraint was removed, and I suspect we may have made more senior appointments than was ideal. It was always difficult to turn down someone who had already established themselves as an outstanding researcher, in favour of someone who was younger, but more of a risk. Of course, this meant that people could not have as much space or as many RAs and post docs as they would have liked. I encouraged people to collaborate with colleagues elsewhere who had more access to space. I don’t regret this, since I think it enhanced the positive influence of the Unit. It did however, contribute to an age structure where quite a number of people were likely to “grow old together”, something that I know Donald was intent on avoiding. Apart from that however, I think our strategy of having a relative flat management structure with a large number of senior people worked reasonably well.

John Fox

For this meeting, John Fox wrote a tribute in memory of Donald Broadbent and Allen Newell, with the title Images of Mind. His abstract runs as follows:
The idea of “mind” did not spring fully formed into human consciousness. On the contrary it has been articulated slowly through the millennia, drawing upon countless metaphors and images in different cultures at different times. In the last 50 years the concepts of conventional science and technology have provided the primary images that we employ in discussion of mental processes, though there are presently many competing perspectives. Every one of these images is incomplete when it comes to explaining mental phenomena, and many are inconsistent. In this chapter we review a few of the prominent perspectives that have influenced cognitive science in the last half century or so, from information processing psychology to AI, and conclude that a unified theory of mind will need insights from multiple viewpoints. The challenge to the field is to avoid disputes over different positions and look for ways of bringing them together.
Keywords: Agents, AI, cognitive systems, knowledge representation, cognitive modelling, mental states, cognitive theory.

The full text of John’s article can be downloaded  Images of mind

Neville Moray: Post-conference comments

1. Donald told me that he believed he had read all the relevant material that had been published anywhere when he wrote “Perception and Communication”, but by the time he wrote “Decision and Stress” that was no longer possible because the progress of research had been so rapid. (I don’t believe he was quite correct, because I don’t think he could read many foreign language papers, but I know what he meant.)

2. He told me during the dinner at his retirement “fest” in Oxford that he had resigned because the political framework within which any Director of APU now had to operate had changed so much that he no longer felt he could handle it. he said that in his early years he knew how to make things happen, how to deal with the scientific establishment, etc., but that things had changed so much he no longer felt confident in being able to make things happen. He also felt that the “democratisation” of the post-1960’s caused so much time to be spent sitting around discussing what to do that he found it very frustrating. In short, he did not find the “new” world congenial to his view of how to run the APU, and he felt he could serve psychology better by giving the job to someone else, and spending the rest of his time as a researcher.

3. Perhaps the nearest equivalent to the APU was the Institute for Perception RTO-TNO in the Netherlands. You may want to contact Andries Sanders (who has now retired) in Holland for his views about APU, and perhaps some of the other people who worked there in the 1950-1970 period. Another comparable institute is the Canadian DCIEM at Toronto.

4. It occurred to me when I had my little set-to with Tony Marcel that he was thinking of a fundamental difference between psychologists and engineers. But to me the main difference is between applied and “pure” research. Applied people, whether engineers or psychologists are concerned primarily with prediction: it does not matter if they know their theories to be incorrect providing they predict what will happen. “Pure” researchers (academic psychologists) are concerned with explanation, not prediction. The APU could just as well have been called the Institute for Ergonomics or the Institute for Human Factors. A related difference is that applied work needs not just significant but large differences between conditions. I think that this was less emphasised by the APU.

5. I have never understood why there were no engineers at APU. In particular the work on motor skills done by Christopher Poulton seems to me to have been really rather primitive compared with the work that was done by engineers in other countries to model the human operator. UK work on skills did not in those days make any use of control engineering, despite the early work of Craik, Tustin, etc..

6. I was interested in the remarks about when a computer was acquired. With Harry Kay I put one into Sheffield in, if I remember correctly, 1987. (We beat Stuart Sutherland at Sussex by 6 months, and were the first university psychology department to have one.)

7. As I said in the meeting, it is salutary to look at the chapters on Skills and Engineering Psychology in the 1953 edition of S. S. Stevens “Handbook of Psychology” to see how enormously advanced the work of the APU was in ushering in the cognitive revolution. It is really amazing.

8. when I began consulting to industry, the military, etc., in the UK I found my academic colleagues were very scared of the idea. “Suppose you get it wrong?”. One importance of the APU is obviously that it led the way in getting industry, commerce, and society at large used to the idea that psychology really could make a difference for the better, and was not to be equated with psychiatry, and especially psychoanalysis.

9. I always felt it odd that the APU was a child of MRC rather than DSIR. I was interested in the remarks about not letting the programme overlap too much with SSRC. Wasn’t Peter Warr’s Unit at Sheffield an MRC unit at least originally, and it was called the “Applied and Social Industrial Psychology Unit” or something like that. Perhaps I’m wrong and it belonged to SSRC. If I’m right, why the difference?

10. In the “Purple Book”, the proceedings of the earlier meeting about the APU, it is said that Donald had a reputation for chewing up people who disagreed with him. If that is true is must have been an “in house” phenomenon. He was extremely kind, generous and helpful to me and to Anne Treisman (then Taylor) when we were working on our D.Phil.s in 1958-59. He responded to brash young attacks on his work by me with great good humour and an impression of taking me very seriously and enjoying the argument. This was always the case during my career, and he behaved in the same way to one of my Ph.D. students who found some of his work unrepeatable. He would raise his eyebrows, put on his “Tony Hancock” expression, and discuss the problems with great good humour and incisive comments. I was very surprised to read the passage in the “Purple Book”.

11. Another thing I think was missed was the enormous importance of cultural differences. The APU like must of us in those days (and many now) thought in terms of “universal” findings. Now we know that there are very important differences in performance which are a function of the culture from which the “subjects” come. Indeed I now recall that once when I obtained a different result from Donald when I repeated some of his work he said that he repeated the experiment (which had originally been done on sailors) on members of the APU or on students and obtained my results. I can’t now remember what the experiment was, but it was one of the split-span memory tasks I think. But the piossibility of cultural differences was not made much of.

12. Someone in the discussion made the point that with applied work there is always uncertainty about what part of the task has to be reported or transferred to future work if the results are to be generalised. There is an excellent discussion of this (by an engineer!) in Rasmussen, J., Pederesen, A-M., & Goodstein, L. (1995). Cognitive Engineering: concepts and applications. New York: Wiley, pp. 219-24.

13. I think one of the most important features of the APU was that it provided the “critical mass” for discussions, whereas in the 1960-1980 period those of us in university departments were lucky if we had one other colleague in roughly the same area.

14. As someone said, a very important feature was the development of new technology. When Anne Treisman and I began our graduate work in 1957 it was the first moment when tape recorders were readily available, and even then we had to have 2-channel recorders manufactured to order by the manufacturers: none were commercially available. Similarly, it is hard to remember that I took the second course in computer programming that was ever given at Oxford, in 1958. It is interesting that Poulton was still using smoked drums in his motor skill work. The technical services were critical for the success of APU.

Michael Posner (who was unable to attend) wrote

A SNAPSHOT OF LIFE AT THE UNIT IN 1968-69 Michael I. Posner, University of Oregon

I was at the APU during the year 1968-1969. My PhD advisor Paul Fitts shared with those at APU an effort to develop theoretically based applications of psychology. He called his approach to this effort Human Peformance Theory and it was very influenced by the developments at the APU . Fitts knew Sir Frederic Bartlett and wrote to him about my dissertation. My own research had been influenced by Donald Broadbent’s 1958 book and I was anxious to spend time at the unit. The opportunity came in 1968, after I had moved to the University of Oregon, thanks to the National Science Foundation I received a senior postdoctoral fellowship that provided funding to spend a year at the Unit.

I had paved the way for the visit in 1966 when, after attending the International Congress, of Psychology in Moscow and the first meeting of Attention and Performance in the Netherlands, I came to Cambridge to see about the years visit. I was very excited about the work going on there and Bob Wilkinson was willing to have me work with him during the following year.

When I was at APU, Donald Broadbent was assembling the material that was to be Decision and Stress. He did several seminars on the topic. The blending of signal detection theory with applied topics like time of day effects and work load was very stimulating. Several of the researchers in the audience had carried out the work Donald discussed and there was both enthusiasm and apprehension for his approach. I thought it was very brave of Donald Broadbent to draw together the work of the Unit into theoretically motivated categories while the people who had performed the research were present to object to his handling of each experimental detail. Nonetheless it was a wonderful lesson in the blending of research and theory and, like his 1958 book, provided me personally with motivation to carry on running experiments in the hopes that someone like Broadbent would weave them into a broader pattern.

I tried to keep up with some of the overall work of the Unit. Many of the ideas I heard there were incorporated into the book Cognition, which, while it was not published until 1973, was largely written during the year I was in Cambridge. However, my main project was with Bob Wilkinson. Grey Walter had recently published his work on the contingent negative variation (CNV), a negative shift in the scalp recorded EEG that appeared during the foreperiod of a reaction time task. His work implied that the CNV took .5 to one second to begin, much too late for it to be responsible for the improved RTs, since they required only a warning of .2-.5 . Our studies showed clearly that the negativity that was to be the CNV could be seen within the first 100 millisec, superimposed on the warning signal ERP. I suppose we were actually doing cognitive neuroscience, tieing together brain studies with cognitive operations, but we did not know it then. We also tried to apply this logic by attempting to determine the relation between phasic changes induced by the warning signal and tonic differences due to time of day. We tested naval ratings at the peak of their diurnal cycle in the early evening compared with their performance at 2-3 AM. However, the computer of average transients that we used to record the EEG and the tachistoscope, needed to show the stimuli, were so complex to operate in combination, that at 2AM I made so many experimenter errors we could not be sure whether warning signal effects and tonic effects of time of day were additive or not.

Although in my case there were no practical recommendations flowing for the studies I performed, it was clear that the combination of the theoretical issues raised in Decision and Stress with the practical demands of the agencies that funded the Unit’s work were a major reason for its overall success. In the lab next to where I was working with Wilkinson there was an effort to develop an prosthetic device for blind persons. While the effort was far from practical application it provided an inspiring potential for many visual experiments taking place at the Unit. The efforts of Pat Wright and others to develop tools for designing and evaluation of the difficulty of documents was another example of how experimental and applications worked together. When Dan Kahneman visited the unit he helped Pat learn how to use pupil size to gage the extent of difficulty a person was having in a cognitive task. I still look at my wife’s pupils when we pour over the internal revenue rules each year at tax time and wonder why Pat’s ideas were not more widely applied.

I was not in a position to evaluate the internal workings of APU in comparison with groups in the US. I do know I had a lot of fun at the APU. Probably a lot of the fun was due to the presence of a number of young investigators who came to the unit the same year I did. I enjoyed almost daily walks with Bob Hockey and Peter Hamilton to the Cambridge union, where we enjoyed a pint and they often stopped for a game of snooker. Other prominent events, were a wonderful Guy Fawkes day spent with Derek Corcoran and his family. I can still feel the warmth of the fire and of their hospitality. Of course the Cambridge environment held its own charms. I was invited to some dinners in the Colleges, one especially with Donald Broadbent comes to mind. I was amazed at the speed at which the servers advanced our dinner from course to course. The complex sound environment made my conversation with the man seated next to me difficult. I thought he said he was a sociologist of the fens, but realized my comment were not apt when I realized it was of the family. Our final days in Cambridge were marked by a dance held at the unit. It was a wild evening and with some drinks under my belt I even tried one of the then popular and rather fast dances. I was not a success! However, the evening was definitely warm wild and wonderful: so much for British reserve!

Willem Wagenaar (who was unable to come) wrote

I began my psychology study in 1960, at the University of Utrecht. Psychology in the Netherlands was at that time oriented towards the German pre-war tradition. Phenomenology was prevailing, Klages and Lersch were the heros, Wundt and Donders were the sources for experimentation. Little was known about what happened in the Anglo-Saxon world. My professor of Experimental Psychology, Johannes Linschoten, did his PhD on Stereopsis, mostly in the phenomenological tradition. He also published a book on William James, which appeared to be a first step in the direction of a more Anglo-Saxon type of experimental psychology.

In 1949 a completely different line of research had been started by the Dutch Defence Ministry, who were interested in problems of night vision. They asked the ophtamologist F.P. Fischer at Utrecht (who was the teacher of my father, and later my father in law!) to look at night vision problems. He advised the Ministry to hire Maarten Bouman, a physicist who had written his PhD about the quanta threshold of vision: two photons hitting one receptor would be enough to enable the detection of light. Bouman started with my father in law and another ophtamologist Jurn ten Doeschate in what was called the Working Group Perception. Soon the group was promoted to The Institute for Perception TNO, which moved to Soesterberg, where it has been ever since. Bouman became the Director. Around 1958, or so, Bouman realised that he needed experimental psychology, and experimental psychologists. But there was no real experimental psychology in The Netherlands. Therefore Bouman took two steps. He went to Cees Sanders, the director of the National Psychological Service (mainly involved in personel selection for government positions). And he consulted Linschoten. Sanders confirmed that there were no real experimental psychologists in The Netherlands, and that there was only one person who could fill the gap: his younger brother Andries who had just completed his masters degree. That is how things go in our country, and see how fortunate this advice was!

Linschoten told Bouman that he should hire a young psychologist, who then needed to be trained in the APU of Broadbent. As a result Andries Sanders was hired by the Institute for Perception. He was sent, as far as I know but you should ask him for the precise details, to Cambridge for a full year, in 1961. There I met him when I was visiting APU as a student. In the following years several more psychologists were hired by Bouman: John Michon, Willem Levelt, Len de Klerk, and me in 1965. John van de Geer became an advisor after Linschoten died in 1964, and the PhD supervisor of about everybody who counted in experimental psychology and datatheory. In this manner soon the entire field of experimental psychology became “Anglo-Saxonized” in the Netherlands, and it came all from one room in the Institute for Perception where a national tradition had been built upon Andries Sanders experiences at APU.