Why can't most people draw what they see?
Cohen DJ, Bennett S · 1997
grade Dquasi-experimentindependentunreplicated
Sample
Four experiments; Experiment 4 had 39 people drawing (28 untrained, median 0 months of art education; 11 advanced art majors, median 12 months) and 156 raters
Population
US college students, both untrained and formally trained in art
Design
Not a training study, and it matters that the archive says so: this is an error-decomposition design that successively removes drawing sub-abilities (tracing, distance judgement, self-assessment) to locate where the error lives. There is no instructional manipulation anywhere in it. Included because it identifies the bottleneck any art curriculum would have to move, and because of one line the authors wrote in 1997 that is still true.
Key findings
The conclusion, verbatim from the abstract: "(a) motor coordination is a very minimal source of drawing inaccuracies, (b) the artist's decision-making process is a relatively minor source of drawing inaccuracies, and (c) the artist's misperception of his or her work is not a source of drawing inaccuracies. These results suggest that the artist's misperception of the object is the major source of drawing errors." The sentence that belongs in this archive is on page 611: "Although there is no experimental evidence that drawing inaccuracies resulting from delusions can be overcome, art students readily acknowledge these errors and correct them." Twenty-eight years later that experimental evidence still does not exist. The only training-related number is descriptive and untested: critics rated trained participants at 2.96 (SD 0.21, 22 drawings) against untrained 2.65 (SD 0.32, 56 drawings), with the authors noting the two distributions overlapped.
Genetic confound
Not addressed. The trained-untrained contrast is cross-sectional and self-selected and the authors do not treat it as an effect of training.
Replication notes
No direct replication located. The perceptual account has been refined by later cross-sectional work (Perdreau and Cavanagh 2014 argue drawing expertise reflects more efficient encoding of object structure rather than more veridical perception), none of it experimental.
Effects
| Outcome | Metric | Value | Measure | Timing | Vs | Horizon | Class |
|---|---|---|---|---|---|---|---|
| Source of drawing inaccuracy, decomposed | qualitative verdict from four experiments | misperception of the OBJECT is the major source; motor coordination is minimal; decision making is minor; misperception of one's own work is not a source | researcher-designed | single session | none | not-applicable | domain-skill |
| Judged drawing skill, formally trained versus untrained participants | critic rating (log scale, lower is more accurate) | trained M = 2.96 (SD 0.21, n = 22 drawings) versus untrained M = 2.65 (SD 0.32, n = 56 drawings); NO significance test was reported for this contrast, and the authors note the skill distributions overlapped | researcher-designed | cross-sectional | none | not-applicable | domain-skill |
Cited by
- Does teaching visual art produce artistic skill?insufficientconf: lowgc: medium