Thursday, September 29, 2011

Development of drawing abilities


The book Making Sense of Children's Drawings by John Willats puts forth a compelling theory of how kids learn to draw, and the course of that development.

To Willats, drawings link up to our mental conceptions of how things look in the world, thereby rejecting a view that says drawings are entirely based on what we see. A great example of this is when children are given dice and told to draw them. Instead of drawing them as they see them, they draw dice with all six sides, which would be impossible to see.

Willats also provides great detail on the origin of the "don't copy" trend of instruction in drawing (which he, like me, is highly critical of). As he describes, this came originally from the 1800s educator Franz Cižek, based on Romantic ideas that children had a pure "inner creativity" that needed to develop unspoiled by imitation from external influences.

As he nicely points out, this doctrine is largely not reflected by what children actually do. Indeed, closer inspection of Cižek's own students show a consistent group style. They were copying between each other, just not from him.

Also, his general trajectory for learning to draw runs like this:
  • 1-3 yrs: dots, lines, regions. Scribbles denote whole regions of space, not necessarily just random uncontrolled lines.
  • 2/3-8 yrs: Bounded areas depict regions and volumes. Round, long regions denote round long volumes, while long or round regions show flat volumes.
  • ~6-10 yrs: Regions are used as as picture primitives to denote faces rather than volumes. However, lines still denote boundaries of regions, not the contours of shapes.
  • ~6-8 yrs: Regions as volumes, but compensated by more modifiers, resulting in "having a smooth outline" (threading); denote regions n the visual field (starts approaching lines as contours)
  • ~8-10 yrs: Finally, lines are used as picture primitives (instead of using lines for regions). Lines are finally used as contours, as evident by line junctions used for occlusion and foreshortening.

The one drawback to this approach is that, despite his critique of the overall trend against copying, his developmental trajectory does not incorporate the effects of imitation on drawing. This may not be possible for him though: there simply doesn't seem to be enough data, looked at through the right perspective, to offer his model much more (true both of when the book came out, and now).

However, overall, I highly recommend this book to anyone interested in how children learn to draw.


Willats, John. 2005. Making Sense of Children's Drawings. Mahwah, NJ: Lawrence Erlbaum.

Tuesday, September 20, 2011

Graphical Abstracts

A friend of mine passed along this interesting link today. The academic publisher Elsevier looks like it's now accepting "graphical abstracts" for scientific papers in journals:

A Graphical Abstract should be a one-image file and should visualize one process or make one point clear. For ease of browsing, the Graphical Abstract should have a clear start and end, preferably "reading" from top to bottom or left to right. Try to reduce distracting and cluttering elements as much as possible.

Also interesting is that they expressly specify how they want the image to be "read." Now I'm curious what they'd think to an abstract using sequential images...

Wednesday, September 14, 2011

What is the human language faculty?

My mentor, Ray Jackendoff, has a new article out in the journal Language that mentions my research (as well as has some illustrations by me):

Jackendoff, Ray. 2011. What is the human language faculty?: Two views. Language 87(3):586-624

The piece explores the biological foundation of our capacity for language, and what components of cognition contribute to language understanding. My work comes in because he points out that several cognitive capacities involve the hierarchic organization of structures, including language, music, vision, events, and, yes, the visual narrative in comics. Here's the abstract:

In addition to providing an account of the empirical facts of language, a theory that aspires to account for language as a biologically based human faculty should seek a graceful integration of linguistic phenomena with what is known about other human cognitive capacities and about the character of brain computation. The present discussion note compares the theoretical stance of biolinguistics (Chomsky 2005, Di Sciullo & Boeckx 2011) with a constraint-based PARALLEL ARCHITECTURE approach to the language faculty (Jackendoff 2002, Culicover & Jackendoff 2005). The issues considered include the necessity of redundancy in the lexicon and the rule system, the ubiquity of recursion in cognition, derivational vs. constraint-based formalisms, the relation between lexical items and grammatical rules, the roles of phonology and semantics in the grammar, the combinatorial character of thought in humans and nonhumans, the interfaces between language, thought, and vision, and the possible course of evolution of the language faculty. In each of these areas, the parallel architecture offers a superior account both of the linguistic facts and of the relation of language to the rest of the mind/brain.

Keywords:
narrow faculty of language, recursion, parallel architecture, Merge, Unification, lexicon, consciousness, evolution

Thursday, September 08, 2011

Segmentations in visual narrative

Gernsbacher's 1985 paper "Surface information loss in comprehension" is an important article on the comprehension of sequential images, and one that has informed much of my current research. It is based on her dissertation, and describes several experiments.

Overall, Gernsbacher had participants read the Mercer Mayer book Frog, where are you? to question whether people can accurately recall the exact surface images in the story, or if (like language) they are only able to retain the gist of meaning.

First, she asked participants to read this "picture story" and choose where they would divide it into parts. They simply drew lines between images where they felt that one episode ended and another began. Overall, she found that people greatly agreed on where these boundaries between segments were placed.

She then asked another group of people to read the stories, but the composition of certain images were flipped horizontally. These images either came before or after the boundaries that people agreed upon in the previous experiment. She found that people had a harder time accurately remembering the horizontal composition if the image came after the boundary as opposed to before it. This provided evidence that people were building up context throughout a segment, and that the start of a new segment incurred a cost on memory.

These experiments were important for several reasons. First, it confirmed her hypothesis that people mostly retain the gist of meaning and not the surface information of images. Given that people's comprehension did not appear overly damaged by flipping the composition of images, it could be pertinent to discussions of how much impact is really made by the left-right composition of images, such as in the 180º rule.

However, more importantly, these experiments showed very strong evidence that people group images together into segments. This poses a problem to theories like McCloud's panel transitions, which  envision no stopping point for linear transitions: they keep going on and on throughout a visual narrative (either linearly or promiscuously between multiple panel relationships).

Rather, this experiment shows that people have some intuitions for dividing up visual narratives into segments (what I called in my book "visual sentences"), and that moving between those segments incurs a cost to comprehension.


ResearchBlogging.org
Gernsbacher, Morton Ann (1985). Surface information loss in comprehension. Cognitive Psychology, 17 (3), 324-363 DOI: 10.1016/0010-0285(85)90012-X

Saturday, September 03, 2011

Waning days of studenthood

Tuesday will begin what looks to be my last year of grad school, which means my last year of being a student. Yikes! That means I'm currently balancing finishing my projects, helping teach classes, writing/revising papers, and looking for what's next. Should be a wild semester!

I have quite a few projects underway right now, starting with my second brainwave study looking at the comprehension of sequential images. I got some good data on this experiment over the summer with a "reaction time" experiment, so now its time to stick electrode caps on people! If all looks good with troubleshooting, we could be up and running this week.

Also, coming up in October I'll be making an appearance down at the New York ComicCon, giving a talk on my research on Saturday, October 15th. As the date gets closer, I'll offer more info.