Investigating the correlation between the intelligence and the performance in repetitive learning tasks

E. J. Gorzelanczyk, P. A. Wozniak, H. Chrzastowska, 3rd Congress of the European Federation of Neurological Societies, Sevilla, 1998

Do intelligent students learn faster?

The article presented below seems to indicate that your IQ does not really matter much in learning with SuperMemo!
It has been shown earlier that the learning process in SuperMemo correlates strongly with your learning results. Strangely, good students showed lower retention as measured by SuperMemo! This can easily be explained by the fact that SuperMemo is based on self-grading. Critical students provide lower grades and thus reduce their retention measurements. However, their high self-criticism pays off when it comes to exams that verify the learned knowledge. The fact that little correlation has been found between the same learning parameters and the IQ and its components, indicate that the motivational factor may really play a much higher role in learning that it is popularly perceived.

The conclusions for you, the user of SuperMemo:

  • Never say: "I am too stupid to learn it". If you cannot understand something, you probably miss some facts needed to complete the picture or perhaps the book you are studying is just badly written. Do not blame yourself! If you really try and you cannot, it probably is not your fault
  • Never say: "I have too bad memory to really learn it". The experience of thousands of students of SuperMemo is that on average we differ very little as far as basic memory properties are concerned! Some students can really better structure the material to learn and thus learn really fast but with SuperMemo ... you will be forced to develop better mnemonic techniques (ways to structure the material in your memory). Otherwise the intractable material will make your repetitions a living hell. With time, you will see that your speed of learning does not differ much from the speed of other students.
  • What really counts most in learning is motivation! If you really want to learn something, the only true limitation is time and the availability of a well-written (or well-structured) learning material. With SuperMemo, your memory no longer is a factor!

The authors have been investigating the applications of repetition spacing algorithms in learning (note that SuperMemo algorithm is a repetition spacing algorithm). As significant variations in the speed of learning have been observed among the tested students, the question has been raised about the possibility of finding a correlation between the components of intelligence and personality and various measures of the learning process in a group of students.

In the presented study, thirteen different measures of the learning process have been juxtaposed against fourteen components of intelligence as measured by the Wechsler test for Polish population. Thirty students participated in the test. Their IQ has been measured and they all have been subject to a learning process based on repetition spacing lasting twelve months.

Earlier studies have shown that such a learning process, when based on self-grading, often shows surprising negative correlation between the measured knowledge retention and final knowledge tests. Better students, i.e. those who fared well in the final test, have consistently showed lowered retention measurements throughout the learning process. In self-grading procedure, it appears that more critical students provide lower grades at repetitions and consequently show lower retention as measured by SuperMemo (i.e. not necessarily lower retention in general); however, their self-criticism and increased repetition workload make them fare better in verification tests.

In the presented study, the authors have been looking for correlates between the parameters of the learning process and various components of intelligence. No significant correlates have been found. This seems to indicate remote relationship between the overall intelligence and the learning process that is probably most visibly affected by motivational factors. Additionally, memory-related components of intelligence do not play a visibly more significant role in learning than other components. This indicates that the basic memory properties of individuals are much less varied in a population than other factors affecting learning such as motivation or persistence.

Description of the measures of the learning process:

Last The date of the latest scheduled repetition after 12 months of learning
1.3s The number of items causing significant difficulties in learning
Burden The measure of the average number of repetitions per day
Grade The average grade given in repetitions (0..5 scale)
Lapses The number of times an average item has been forgotten
Cases The total number of repetitions made in the entire learning period
Retention The average retention throughout the learning process
Day/Rep The average interval between repetitions of an average item
Interval The average interval between repetitions at the moment of completing the process of repetitions
Factor The average difficulty of items as measured at repetitions
Rep The average number of repetitions afforded an average item in the learning period
Workload The average time spent on repetitions per day
Time The average time spent for a single repetition

Learning-Intelligence correlation coefficients (absolute values).

All rows and columns have been sorted for the highest averages (best correlates at the top-left).

Highest correlation has been found for the following pairs:

No absolute value of any single correlation coefficient has passed the 0.5 barrier. Most notably, the correlation between the overall IQ and the retention was nearly zero!

test/parameter

Last

1.3s

Burden

Grade

Lapses

Cases

Retention

Day/Rep

Interval

Factor

Rep

Workload

Time

Average

Maximum

similarities

0.251435

0.115108

0.186151

0.357474

0.192339

0.044872

0.217379

0.446996

0.33504

0.02327

0.126113

0.096549

0.292984

0.219493

0.446996

dictionary

0.19587

0.447874

0.085428

0.201719

0.274271

0.3069

0.261952

0.245754

0.109595

0.360766

0.129229

0.010174

0.041797

0.196172

0.447874

ordering pictures

0.459928

0.007164

0.280348

0.355465

0.104795

0.0405

0.101342

0.325604

0.232347

0.172744

0.098151

0.19341

0.152299

0.191012

0.459928

understanding

0.158112

0.32479

0.1063

0.219757

0.237604

0.255267

0.205431

0.0492

0.10988

0.219496

0.179659

0.148656

0.083561

0.1575

0.32479

math

0.158216

0.220544

0.003081

0.067041

0.293845

0.280389

0.172424

0.093783

0.138667

0.139241

0.064291

0.112015

0.189514

0.150478

0.293845

repeating digits

0.118939

0.03909

0.260202

0.08776

0.098316

0.034617

0.255651

0.050837

0.252497

0.11774

0.0698

0.296262

0.119755

0.134654

0.296262

knowledge

0.303862

0.243793

0.080728

0.048178

0.143129

0.203255

0.161296

0.182687

0.119038

0.046969

0.076874

0.039605

0.12319

0.132473

0.303862

missing pictures

0.246656

0.087238

0.285121

0.239839

0.053011

0.02366

0.01551

0.209288

0.159637

0.026145

0.133331

0.210475

0.01678

0.131344

0.285121

non-verbal scale

0.287654

0.126439

0.2887

0.242132

0.079844

0.056562

0.008464

0.03229

0.072185

0.058815

0.127781

0.206032

0.047984

0.118959

0.2887

verbal scale

0.068362

0.263983

0.14222

0.089241

0.145248

0.10665

0.267287

0.006614

0.0153

0.198578

0.032647

0.063951

0.18209

0.118753

0.267287

IQ

0.042247

0.125142

0.171178

0.065187

0.158327

0.147416

0.025154

0.065028

0.029696

0.156907

0.123413

0.201409

0.230099

0.11278

0.230099

digit symbols

0.005454

0.198295

0.003

0.06796

0.014984

0.176467

0.088343

0.154987

0.188677

0.118531

0.343025

0.00761

0.016567

0.10832

0.343025

bricks

0.212143

0.000299

0.106088

0.040137

0.04751

0.139224

0.066777

0.000828

0.003613

0.105972

0.16183

0.010296

0.099978

0.09095

0.331007

puzzles

0.007805

0.051645

0.17159

0.065381

0.103362

0.114531

0.052283

0.013961

0.054225

0.041374

0.110118

0.164811

0.056728

0.076434

0.17159

Average coefficient

0.179763

0.160815

0.15501

0.153376

0.139042

0.137879

0.135664

0.134133

0.130028

0.127611

0.126876

0.125804

0.118095

  

0.320742

Maximum correlation

0.459928

0.447874

0.2887

0.357474

0.293845

0.3069

0.267287

0.446996

0.33504

0.360766

0.343025

0.296262

0.292984

0.219493

0.459928

For more on the same subject see: Analysis of data collected from a group of students using an algorithm for spaced repetition