Epistemic Spot Check: The Role of Deliberate Practice in the Acquisition of Expert Performance

post by Elizabeth (pktechgirl) · 2019-06-25T23:00:00.689Z · LW · GW · 9 comments

Contents

  “When individuals, especially children, start practicing in a given domain, the amount of practice is an hour or less per day”
     Generalizations about talent development, Bloom (1985)
  “…laboratory studies of extended practice limit practice to about 1 hr for 3-5 days a week”
    Skill and working memory, Chase & Ericsson (1982)
    Controlled and automatic human information processing: I. Detection, search, and attention. Schneider, W., & Shiffrin, R. M. (1977)
    Discrimination reaction time for a 1,023-alternative task, Seibel, R. (1963)
  “These studies show essentially no benefit from durations exceeding 4 hr per day and reduced benefits from practice exceeding 2 hr”
    Fundamentals of Skill, Welford (1968)
    Experimental Psychology, Woodworth & Schlosberg (1954)
  Many studies of the acquisition of typing skill (Baddeley & Longman, 1978; Dvorak et al.. 1936) and other perceptual motor skills (Henshaw & Holman, 1930) indicate that the effective duration of deliberate practice may be closer to 1 hr per day
    The Influence of Length and Frequency of Training Session on the Rate of Learning to Type, Baddeley & Longman (1978)
    Typewriting behavior; psychology applied to teaching and learning typewriting, Dvorak et al (1936)
    The Role of Practice in Fact Retrieval, Pirolli & Anderson (1985)
  Conclusion
None
9 comments

Epistemic spot checks typically consist of references from a book, selected by my interest level, checked against either the book’s source or my own research. This one is a little different that I’m focusing on a single paragraph in a single paper. Specifically as part of a larger review I read Ericsson, Krampe, and Tesch-Römer’s 1993 paper, The Role of Deliberate Practice in the Acquisition of Expert Performance (PDF), in an attempt to gain information about how long human beings can productivity do thought work over a time period.

This paper is important because if you ask people how much thought work can be done in a day, if they have an answer and a citation at all, it will be “4 hours a day” and “Cal Newport’s Deep Work“. The Ericsson paper is in turn Newport’s source. So to the extent people’s beliefs are based on anything, they’re based on this paper.

In fact I’m not even reviewing the whole paper, just this one relevant paragraph: 

When individuals, especially children, start practicing in a given domain, the amount of practice is an hour or less per day (Bloom, 1985b). Similarly, laboratory studies of extended practice limit practice to about 1 hr for 3-5 days a week (e.g., Chase & Ericsson, 1982; Schneider & Shiffrin, 1977; Seibel, 1963). A number of training studies in real life have compared the efficiency of practice durations ranging from 1 -8 hr per day. These studies show essentially no benefit from durations exceeding 4 hr per day and reduced benefits from practice exceeding 2 hr (Welford, 1968; Woodworth & Schlosberg, 1954). Many studies of the acquisition of typing skill (Baddeley & Longman, 1978; Dvorak et al.. 1936) and other perceptual motor skills (Henshaw & Holman, 1930) indicate that the effective duration of deliberate practice may be closer to 1 hr per day. Pirolli and J. R. Anderson (1985) found no increased learning from doubling the number of training trials per session in their extended training study. The findings of these studies can be generalized to situations in which training is extended over long periods of time such as weeks, months, and years

Let’s go through each sentence in order. I’ve used each quote as a section header, with the citations underneath it in bold.

“When individuals, especially children, start practicing in a given domain, the amount of practice is an hour or less per day”

 Generalizations about talent development, Bloom (1985)

“Typically the initial lessons were given in swimming and piano for about an hour each week, while the mathematics was taught about four hours each week…In addition some learning tasks (or homework) were assigned to be practiced and perfected before the next lesson.” (p513)

“…[D]uring the week the [piano] teacher expected the child to practice about an hour a day.” with descriptions of practice but no quantification given for swimming and math (p515).

The quote seems to me to be a simplification. “Expected an hour a day” is not the same as “did practice an hour or less per day.”

“…laboratory studies of extended practice limit practice to about 1 hr for 3-5 days a week”

Skill and working memory, Chase & Ericsson (1982)

This study focused strictly on memorizing digits, which I don’t consider to be that close to thought work.

Controlled and automatic human information processing: I. Detection, search, and attention. Schneider, W., & Shiffrin, R. M. (1977)

This study had 8 people in it and was essentially an identification and reaction time trial.

Discrimination reaction time for a 1,023-alternative task, Seibel, R. (1963)

3 subjects. This was a reaction time test, not thought work. No mention of duration studying.

 

“These studies show essentially no benefit from durations exceeding 4 hr per day and reduced benefits from practice exceeding 2 hr”

Fundamentals of Skill, Welford (1968)

In a book with no page number given, I skipped this one.

Experimental Psychology, Woodworth & Schlosberg (1954)

This too is a book with no page number, but it was available online (thanks, archive.org) and I made an educated guess that the relevant chapter was “Economy in Learning and Performance”. Most of this chapter focused on recitation, which I don’t consider sufficiently relevant.

p800: “Almost any book on applied psychology will tell you that the hourly work output is higher in an eight-hour day than a ten-hour day.”(no source)

Offers this graph as demonstration that only monotonous work has diminishing returns.

 

p812: An interesting army study showing that students given telegraphy training for 4 hours/day  (and spending 4 on other topics) learned as much as students studying 7 hours/day. This one seems genuinely relevant, although not enough to tell us where peak performance lies, just that four hours are better than seven. Additionally, the students weren’t loafing around for the excess three hours: they were learning other things. So this is about how long you can study a particular subject, not total learning capacity in a day.

Many studies of the acquisition of typing skill (Baddeley & Longman, 1978; Dvorak et al.. 1936) and other perceptual motor skills (Henshaw & Holman, 1930) indicate that the effective duration of deliberate practice may be closer to 1 hr per day

The Influence of Length and Frequency of Training Session on the Rate of Learning to Type, Baddeley & Longman (1978)

“Four groups of postmen were trained to type alpha-numeric code material using a conventional typewriter keyboard. Training was based on sessions lasting for one or two hours occurring once or twice per day. Learning was most efficient in the group given one session of one hour per day, and least efficient in the group trained for two 2-hour sessions. Retention was tested after one, three or nine months, and indicated a loss in speed of about 30%. Again the group trained for two daily sessions of two hours performed most poorly.It is suggested that where operationally feasible, keyboard training should be distributed over time rather than massed”

 

Typewriting behavior; psychology applied to teaching and learning typewriting, Dvorak et al (1936)

Inaccessible book.

The Role of Practice in Fact Retrieval, Pirolli & Anderson (1985)

“We found that fact retrieval speeds up as a power function of days of practice but that the number of daily repetitions beyond four produced little or no impact on reaction time”

Conclusion

Many of the studies were criminally small, and typically focused on singular, monotonous tasks like responding to patterns of light or memorizing digits.  The precision of these studies is greatly exaggerated. There’s no reason to believe Ericsson, Krampe, and Tesch-Römer’s conclusion that the correct number of hours for deliberate practice is 3.5, much less the commonly repeated factoid that humans can do good work for 4 hours/day.

 

[This post supported by Patreon].

9 comments

Comments sorted by top scores.

comment by waveman · 2019-06-26T11:12:18.056Z · LW(p) · GW(p)

There seem to have been a few individuals who could work hard for more than 4 hours a day: Proust (who took vast amounts of caffeine tablets and died at 51), Erdos (who used amphetamines), Richard Stallman who was and is a super motivated individual.

In the book daily rituals, about high achievers, few worked more than 4 hours on their core hard work e.g. writing novels, science etc. You would think if it were possible to work productively at the top level more than that, someone would do it and blow away the competition.

I would be interested in any others, or any evidence that people in general can do more than 4 hours at the top level. Possibly a nap after 3 hours can get you another 30-60 minutes. This was from the violinists study that Cal Newport (I think) referred to.

In general people tend to initially find the 4 hour limit a big problem. My response is to ask people to get back to me when they are consistently doing the 4 hours and we will see how it can be extended. They tend to find it is very hard to get to 4 hours.



Replies from: pktechgirl, An1lam
comment by Elizabeth (pktechgirl) · 2019-06-26T19:51:51.073Z · LW(p) · GW(p)

is there a reason you're anchoring at 4 hours in particular?

Replies from: waveman
comment by waveman · 2019-08-07T01:41:57.790Z · LW(p) · GW(p)

Why 4 hours?

1. 4 hours a day has been widely reported as the limit

2. The book Daily Rituals reports high achievers doing 4 hours really hard work a day.

3. Personal experience. Steep drop off after more than 4 hours; burnout after a few days much over 4 hours, etc.

4. Very few examples of people going over that number sustainably.

I suggest people track this themselves and see what happens.

I find I can get to 4.5-5 hours maybe with a lunchtime nap. Maybe much more with lots of micro-naps (doze in chair for 5 minutes).

Currently I am experimenting with turning 24 hours into two days with a long nap in the middle. I am having trouble doing this though.

N.B. This is not 4 hours of any kind of work. This is work at the maximum of intellectual effort e.g. deliberate practice, learning to ride a bicycle, memorizing vocabulary with Anki decks, practising a foreign language at the limit of your comprehension, trying to prove theorems, doing exercises on a hard scientific subject you are learning, writing at the top level of quality and/or on difficult topics, etc.


Replies from: waveman
comment by waveman · 2020-02-13T05:58:29.807Z · LW(p) · GW(p)

Having said all that, there is a crying need for more work in this area.

The current lead I am following up is Herbert Simon. Will also check out Knuth.

Someone suggested Flaubert, who worked 12 hours a day. And produced 0.7 (really well honed) words per hour.

Replies from: waveman
comment by waveman · 2020-02-16T01:03:48.543Z · LW(p) · GW(p)

Follow up on Herbert Simon. From his book "Models of my life"

Worked 60-80 hours a week. But does not detail what "work" means.

When collaborating with someone he comments that most of his day's work would be usually done by 10am, about the time his collaborator would be getting started. This perhaps hints that early in the day he did a few hours of really hard intellectual work.

What HS regarded as hard work may differ from other people. For example he learned about 20 languages to the point of being able to read papers, and 4-5 to the level of reading literature. But he regarded this as a fun/hobby thing.

He had a problem of hobbies turning into work, and had to drop several of them (e.g. playing musical instruments).

At college he only did enough work to get graded at A. Early on he spent too much time playing ping-pong and his grades slipped.

He published ~1,000 papers and 37 books and accrued to date over 350,000 citations. So he was amazingly productive.

He spent a lot of time on office politics and other managerial and administrative things.

He found writing easy and so wrote many/most of the papers he was a collaborator on.

Conclusion: HS was very smart, very productive, found things that were challenging for others to be fun/hobbies, and while it seems he did work long hours, it is not clear how much time he spent at the highest level of effort. There are hints he did concentrate his top tier work in the first few hours of the day.

comment by NaiveTortoise (An1lam) · 2019-11-03T13:26:59.091Z · LW(p) · GW(p)

In this interview, Don Knuth gives the strong impression that he works more than 4 hours a day not necessarily doing deliberate practice but definitely hard cognitive work (writing a book that most people consider quite challenging to read). That said, Knuth is kind of a monster in general in terms of combining really high technical ability and really high conscientiousness, so it wouldn't surprise me if he's similar to the other outliers you mentioned and is not representative.

A few examples to back up my conscientiousness point:

  • The following story about doing all the problems in the textbook (from that interview):

But Thomas’s Calculus would have the text, then would have problems, and our teacher would assign, say, the even numbered problems, or something like that. I would also do the odd numbered problems. In the back of Thomas’s book he had supplementary problems, the teacher didn’t assign the supplementary problems; I worked the supplementary problems. I was, you know, I was scared I wouldn’t learn calculus, so I worked hard on it, and it turned out that of course it took me longer to solve all these problems than the kids who were only working on what was assigned, at first.

  • Writing an entire compiler on his own over a summer.
  • Finishing his PhD in three years while also consulting with Burroughs.
comment by gwern · 2019-08-09T20:21:16.588Z · LW(p) · GW(p)

Fundamentals of Skill, Welford (1968)

I've uploaded a scan if you want to look.

comment by Alex_Altair · 2022-10-02T20:56:40.093Z · LW(p) · GW(p)

The Role of Deliberate Practice in the Acquisition of Expert Performance (PDF)

This link seems broken (though a google search finds many copies of the PDF).

comment by sirjackholland · 2019-06-27T14:49:53.814Z · LW(p) · GW(p)

Thanks for the spot check! I had heard this number (~4 hours per day) as well and I now have much less confidence in it. That most of the cited studies focus on memorization / rote learning seriously limits their generality.

Anecdotally, I have observed soft limits for the amount of "good work" I can do per day. In particular, I can do good work for several hours in a day but - somewhat mysteriously - I find it more difficult to do even a couple hours of good work the next day. I say "mysteriously" because sometimes the lethargy manifests itself in odd ways but the end result is always less productivity. My folk theory-ish explanation is that I have some amount of "good work" resources that only gradually replenish, but I have no idea what the actual mechanism might be and my understanding is that ego depletion has not survived the replication crisis, so I'm not very confident in this.