Why Bloom’s Taxonomy Makes No Sense

It’s always interesting to see, hear and read the buy-in on Bloom’s Taxonomy, and the grand ascriptions to its rather dull precepts.

There are plenty of academic papers on it, and it’s often the theoretic basis for research, theses, dissertations, discipline-centric and state DoE accreditations. It’s also the most oft-cited framework for the alignment of instructional activities to the 6 identified (version-specific) cognitive acts (about 380,000 results came up in 0.25 seconds in a Google search just now) in teaching and training contexts.

Despite my 40 years of groping for meaningful research, there have been few studies correlating the efficacy of aligning instructional tasks to the levels of Bloom’s hierarchy with the gold standard—student outcomes, and these are dubious at best.

I sought Bloom out in the late 1990s, because he’d made me a convert to the mastery learning approach and concomitant assessments, which he invented and championed (with credit to inspirations of the Winnetka Plan). Yet, in personal conversation, he likened his Taxonomy and the process of its application to “trying to hold onto a freshly caught salmon—the harder you grab onto it, the more it slips away.”

This reminded me of people who faithfully attend church but never take time to read the bible. At least the latter is treated as a definitive text, for which no future editions were planned.

Unlike the bible, there’s been a post-cognitive revolution “millennial” update to the pre-cog 1950s schema, given below.


See http://thecenter.spps.org/uploads/new_and_old_bloom_s.pdf for a short history on this. Surprizingly, there’s been universal buy-in to the update, which I’d expected to engender at least some backlash akin to a religious war. Curiously, there are more Google results (about 496,000 results (0.60 seconds) just now) in the hardly banal philosophic disputes over “early” and “late” Heidegger.

The critiques are few and far between. Most are benevolent (http://www.steve-wheeler.co.uk/2012/06/bloom-and-bust.html) or constructive (http://www.learningspy.co.uk/myths/challenging-blooms-taxonomy).

Bloom was an old man in our first and last encounter, extremely inflexible, a little big on himself, indignant when challenged, and I’m certain he would not have approved of the change. For one, it put 40 years of purported educational research on shaky ground.

Ironically, in the course of decades of driving to schools in New York and Connecticut in order to perform observations for pre-service and permanent teacher certifications, I’ve never seen Bloom’s Taxonomy (and its variants) used in a lesson plan. I have seen teachers reference it in the prefatory language of the unit plans they are required to have on file in the central office, but only when it was mandatory to do so.

A Useful Fiction

I refer to Bloom’s Taxonomy as a useful fiction, because I think broad categorizations of mental tasks (Remembering, Understanding, Applying, Analyzing, Evaluating and Creating) can and ought to be applied in the crafting of behavioral objectives (or the more modern–politically correct–learning objectives).

In that spirit, taking exception to the “low” and “high” level range indications at the top, here’s a more useful rendering.


What I like about the “Teacher Planning” approach is the lists of action words—irrespective of Bloomian categorizations—that practitioners can use to formulate learning activities. The scuttlebut is that teachers hate writing learning outcomes, but we all know that a subject matter expert has these internalized, and they are invariably—implicitly—entailed in the homework we assign.

If a side-effect is the inculcation of knowledge corresponding to the strata of Bloom’s hierarchy—so be it. The fact is expert neural networks are wired differently from novice networks, and it takes a series of conditioning stimuli-response-feedback loops or thought provoking teachable-moments to burn-in declarative and procedural understandings.

Suggestions for Future Research

This motivated me to undertake a study in which I propose to present 5 artifacts of student work in STEAM related subject domains. I would then ask participants to checkmark the corresponding mental acts in Bloom’s Hierarchy (in the guise of an alignment matrix) entailed in the production.

As an extrinsic motivator, I’d offer the choice of having a by-line in the resulting article—a publication credit for academics -OR- a packet of plain or peanut M&Ms for non-academics.

My hypothesis is that every participant will generate a unique covering of the criteria. I wager that it is impossible to look at an artifact of learning, like a smiley-face drawn for an art assignment (the “A” in STEAM) and surmise the mental acts or inspirations behind it. Some kids are just gifted from birth (look at the masterworks Picasso did when he was 10). I also have a colleague who writes poetry by closing her eyes and touch-typing as words come into thought like some divinely inspired haiku. There is no evidence of a propagation through the neural equivalent of Bloom’s schema.

Most of us only arrive at those sorts of “high” level of proficiency through deliberate practice. Either way, the coverings would be idiosyncratic.

It also bothers me that most renditions of the taxonomy are hierarchical, with the mental act of remembering at the bottom and creating at the top.


Maybe we can rank-order the energy expenditures corresponding to these acts in that way, but the arrangement of interstitial acts in the middling ranges seems arbitrary. Moreover, the narrowing of the hierarchy seems to imply less, rather than more effort in “climbing up” to the Zen-state of creative enlightenment.

Perhaps the more accurate rendering of the schema as an inverted pyramid would allude to its tippiness.

Another study would entail looking at the lovely 3D rendering of Bloom’s revised Taxonomy presented below.


I would challenge students to drum up an assignment entailing two or more of the mental acts in that rendering apropos to their content area, aligned to the ordinates of knowledge type and cognitive process. You can kind of do it for one task (for example, as a test of {Remembering, Declarative} list two contenders for the upcoming presidential election) but it gets harder to do as you move up the hierarchy with more substrates to cover.

It’s a tedious chore, at best. In fact, no one really designs instructional activities that way. I’d wager that content experts use a much more top-down than bottom-up approach. Knowing their subjects so well, they simply use the language of their discipline with Bloomean predicates to invoke learning acts in the guise of “assignments.” The purported stages of synthesis are synthetic.

The Undifferentiated Muddle

Having served on two NEASC accreditation committees, written the report, and participated in two NCATE processes, I’ve observed that the same thing goes on in curriculum/standards alignment matrices. I’ve caught administrators red-handed bluffing their way through the requirement by arbitrarily checking boxes in the array to show coverings.They’re usually smart enough to randomize the distribution (unlike kids who get out of the SATs early by drawing Christmas trees with dot fill-ins on the scoring sheet). One administrator flat-out told me it was all “smoke and mirrors” to satisfy external examiners.

You can obtain samples of NEASC and NCATE reports off the websites of these agencies to convince yourself. Good fodder for a dissertation or thesis.

At the very least we could put it in the Suggestions for Further Research section of the proposed study.

The same thing goes on in applications of the Anglo-American Cataloguing Rules (AACR) in the classification of library resources. Ask any two library science students to classify an object, and you’re likely to get different outcomes. Since the 1972 mandate on MARC formats and the evolution of OCRC-like technologies, the cataloging problem has been resolved. Better living through technology.

Continuing on in that domain, there are disparities in the non-contemporaneous generation of Dewey and LC accession codes. The Dewey classification of dogs and cats as technologies is a well-known example (at least among librarians):


Perhaps in the 1890s it made sense to put dog and cat husbandry under agriculture and related technologies, but in our time these notions have devolved into absurdity. The moral of the story is that attempts at consigning subject domains using nominal instantiations is problematic at best, and subject to erosion in the evolution of the vernacular.

The dogs and cats counter-example is cited in a Pew Foundation study (http://www.pewinternet.org/2007/01/31/tagging) that suggests tagging using vernacular terminology as the way to go. This requires the maintenance of a distributed database of vogue usages to keep pace with the morphology of language. Metatagging revolutionized Internet search, manifests in the tagging of friends in photos on Flickr & Facebook and generates billions in revenue for Google.

The analog of branding of mental acts in learning and cognition using schemes like Bloom’s Taxonomy is thus untenable. The demonstrated need for a revision of the very taxonomy itself is symptomatic of its relativism.

Coming Full Circle

There are a few hints as to what’s transpired in the discordant literature since Bloom’s Taxonomy hit the waves.

See http://www.corwin.com/upm-data/13602_Chapter_1_Marzano_Final_Pdf_2.pdf.

As pointed out in that piece, Gagne’s conditions of learning has actually made it’s way far more into contemporary pedagogical practice than the lip service given to Bloom’s taxonomy.

Ironically, Bloom’s “Two Sigma” meta-analysis (http://web.mit.edu/5.95/readings/bloom-two-sigma.pdf) has had impact of late in the evolution of intelligent tutoring systems—that truly are revolutionizing education (https://www.assistments.org/staticpages/Testimonials.htm) and measurably improving student learning.

Yet, Benjamin Bloom is less remembered for that, and Robert Gagne lesser still.

Comments are closed.