Thinking about the boundaries of a revised Bloom taxonomy.

At SIGCSE 2012 I was introduced to a revision of Bloom’s taxonomy of cognitive learning The revision was proposed by L. W. Anderson, D. R. Krathwohl, P. W. Airasian, K. A. Cruikshank, R. E. Mayer, P. R. Pintrich, J. Raths, and M. C. Wittrock in 2001. . I was glad to learn of the revision, for the original had never sat well with me. The new one is a two-axis model, one describing the knowledge and the other cognitive process. Knowledge is categorized as factual, conceptual, procedural, or meta-cognitive; process is categorized as remember, understand, apply, analyze, evaluate, or create. For a gentle introduction to details, see this site.

This taxonomy was raised at SIGCSE by Renate Thies as part of her analysis of CS Unplugged activities in the classroom. Her conclusion was that CS Unplugged is suited for outreach but not for the basis of an educational program because it fails to cover higher categories of knowledge and cognitive processes. This observation has caused me to re-think much of my teaching activities, for quite honestly I think little of my efforts would rank much higher than CS Unplugged. And in that reflection I’ve also started to feel like the revised taxonomy is lacking.

The problems I see are along the knowledge domain. Facts are facts and belong at the bottom of the axis where they were placed. But procedural and conceptual information, while requiring knowledge of facts, do not depend on one another. I see students all the time that display mastery of procedures but lack the conceptual knowledge to know when to apply them. This was true when I was a maths tutor and people would throw one trick after another at a problem until one stuck; it shows up in writing when people un-split infinitives that weren’t actually infinitives at all; it plagues me in CS1 with students who add loops and conditionals to programs that really don’t need either one simply because they know how to write loops and conditionals.

It seems to me that is it beneficial to discuss the factual, conceptual, and metacognitive knowledge of processes. Process vs truth might more appropriately be a third axis, or separate topics to be analyzed within the taxonomy independently if both present in course objectives.

The other problem I have is with lumping so much into the “‍metacognitive‍” category. The examples of the metacognitive category of the taxonomy I can find seem to focus on introspective metacognition: I might analyze my own biases or identify how I remember information. I find it ironic that the pinacle of the taxonomy created by educational researchers is to “‍create introspective cognition‍”—that is, to designing a learning portfolio. But surely introspection, while valuable, is fairly distinct from, say, thinking about the boundaries of conceptual models and postulating new ones. I see no obvious reason why exercising one metacognitive task would strengthen others.

Thinking about thinking is always somewhat confusing, and I am sure we could come up with myriad categories to add if we wanted. One of the keys to creating a good taxonomy is to create it with an objective in mind. The taxonomy presented here may be a good one for most educators: if you can cover the entire spectrum it presents, you have given your students a good spread of knowledge. The extra room at the top of the knowledge hierarchy is probably outside of most educators’ scope. As I said earlier, I have found the taxonomy helpful on a practical level. It hasn’t failed me yet.

There is always something unsettling about anything that tries to exhaustively describe thought. Thinking is so flexible, any taxonomy is doomed to be either vague or incomplete. On the whole, Anderson and Krathwol’s taxonomy seems to me to do a pretty good job of finding a good balance between these.