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

Link post

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].