Last changed 8 Feb 2016   ..............   Length about 5,000 words (57,000 bytes).
(Document started on 15 Mar 2005.) This is a WWW document maintained by Steve Draper, installed at You may copy it. How to refer to it.

Web site logical path: [] [~steve] [best] [this page]


Taxonomies of learning aims and objectives:
Bloom, neoBloom, and criticisms

By Steve Draper,   Department of Psychology,   University of Glasgow.

Bloom's taxonomy originated in an attempt to make assessment more systematic, though it is expressed as being about different types of learning objectives. It is particularly useful, as intended, for help in designing tests e.g. MCQs (multiple choice questions), since we tend to assume that MCQs can only test rote learning, but with care you can test much "higher" kinds of learning. My own interest is in learning and teaching in HE (Higher Education). The commentaries at the end probably show that focus, and may not have the same force in relation to school (K12).

Bloom's taxonomies have since been revised, above all by Anderson, Krathwohl et al. (hereafter "A&K"), and this page's first aim is to summarise those revisions for reference. If you are going to use Bloom's taxonomy today, you need to be aware of, and probably to use, the revised ones. However there is naturally the question of how useful such an old idea (more than 50 years old) still is, so the second aim here is to sketch an evaluation of that.

N.B. The taxonomy is what Bloom is most often cited for; however more important for improving learning and teaching in HE is probably his Mastery Learning work.

Part A: Summaries, pictures, references of Bloom's taxonomy and revisions to it

Diagrams of Bloom's taxonomy

Diagram of Bloom Diagram of Bloom Diagram of Bloom Diagram of Bloom Diagram of Bloom Diagram of Bloom diagram Diagram of Bloom Diagram of neoBloom Diagram of Bloom

Other web pages

  • A version of Bloom:
  • A version:
  • A modification: Now vanished.
  • Improved Bloom, by Anderson & Krathwohl, see Doceo website:
  • Multiple intelligences and Bloom's taxonomy

    Yet other pages

  • GU Guide linking Bloom levels to verbs for use in assessment definitions.

    Tables of neo-Bloom (Anderson & Krathwohl -- A&K)

    Here are tables representing the neo-Bloom taxonomy and its relationship with the original Bloom taxonomy. Some explanations appear after the tables.

    1) Multi-dimensional Bloom: domains of knowledge:
    (Levels X domains)
    level Old (Bloom) Cognitive (new, A&K) Affective Affective2 (Norman Reid) Psycho-motor Psychomotor2 (e.g. dance) Perceptual (me)
    6 Evaluation Creating - World view - Communicative movement Create complex perceptions in others
    5 Synthesis Evaluating Characterising by value or value concept Values Naturalisation Skilled moves Complex scene understanding
    4 Analysis Analysing Organising and conceptualising - Articulation Physical abilities Connoisseurship, classify perceptions
    3 Application Applying Valuing Attitudes Precision Perceptual abilities Active perceptual skills
    2 Comprehension Understanding Responding Beliefs Manipulation Fundamental movements Recognise novel cases
    1 Knowledge Remembering Receiving Knowledge Imitation Reflexes Recognise standard cases
    0 Understanding language Understanding language Feeling - Moving Moving Sensing

    2) Multi-dimensional neo-Bloom: types of knowledge
    [Levels X types, for cognitive domain only] neoBloom scheme in old Bloom table format.
    level Processes (internal, mental) Factual Conceptual Procedural (external behaviour) Metacognitive
    6 Creating Combine Plan Compose Actualise
    5 Evaluating Rank Assess Conclude Action
    4 Analysing Order Explain Differentiate Achieve
    3 Applying Classify Experiment Calculate Construct
    2 Understanding Summarise Interpret Predict Execute
    1 Remembering List Describe Tabulate Appropriate use

    3) A&K usual presentation
    of the 2 dimensions as a blank table to use in analysing a course
    Cognitive process dimension
    Increasing mental complexity of processing →
    Knowledge (type) dimension:
    increasing abstractness

    4) A&K example use
    The definitions of each objective etc. for the course being analysed are normally given in footnotes or a key to the table.
    Cognitive process dimension
    Increasing mental complexity of processing →
    Knowledge (type) dimension:
    increasing abstractness
    Factual Objective1,
    Conceptual   Activity1        
    Procedural   Activity3 Objective3,
    Metacognitive   Objective2,

    5) A&K filled with verb-noun phrases
    Level: 1 2 3 4 5 6
    Processes (internal, mental): Remembering Understanding Applying Analysing Evaluating Creating
    Factual Remember facts Understand facts Apply facts Analyse facts Evaluate facts Create facts
    Conceptual Remember concepts Understand concepts Apply concepts Analyse concepts Evaluate concepts Create concepts
    Procedural (external behaviour) Remember procedures Understand procedures Apply procedures Analyse procedures Evaluate procedures Create procedures
    Metacognitive Remember metacognitive items Understand metacognitive items Apply metacognitive items Analyse metacognitive items Evaluate metacognitive items Create metacognitive items

    6) A&K filled with keywords invented by others
    This is a transposed version of table 2
    Level: 1 2 3 4 5 6
    Processes (internal, mental): Remembering Understanding Applying Analysing Evaluating Creating
    Factual List Summarise Classify Order Rank Combine
    Conceptual Describe Interpret Experiment Explain Assess Plan
    Procedural (external behaviour) Tabulate Predict Calculate Differentiate Conclude Compose
    Metacognitive Appropriate use Execute Construct Achieve Action Actualise

    7) A second rendering of the same table (with colours)

    8) Definitions of processes or levels?
    -- my rationale / restatement of the levels i.e. of the first dimension
    N.B. A&K call the set of 6 levels the dimension of cognitive processes
    level Processes, internal
    A&K redefinition of Bloom levels as internal mental processes
    Procedural, external
    Examples of A&K's Procedural knowledge type (external behaviour)
    New definition (me)
    7 - - Create new categories, concepts, or rules
    6 Creating Compose Generate a new complex case from old elements
    5 Evaluating Conclude Weighted overall judgement of several given complex cases, relating elements
    4 Analysing Differentiate Break down a given complex case into known elements
    3 Applying Calculate Apply old knowledge to new cases (transfer across cases)
    2 Understanding Predict Paraphrase (transfer across descriptions)
    1 Remembering Tabulate Name, recall, and reproduce (elements, relationships)
    0 - - Understanding language

    Some explanatory points

    Graphical orientation

    Bloom's 6 levels are usually depicted as a vertical list, with the higher (more complex) levels at the top; as in the various pictures at the top of this web document. Table 1 also shows the levels in this orientation.

    A&K however show the 6 levels horizontally, with the more complex to the right. Table 3 shows them this way.

    Table 2 is intermediate: it shows the basic A&K 2-D scheme, but in the old orientation.

    The 3 dimensions

    A&K revision of the levels

    A&K slightly relabelled and reordered (swapping levels 5 and 6) the Bloom levels. The levels are shown in the old Bloom order in table 1, with the A&K relabelling also shown (in col.3) (and in table 8 in the new A&K order, with columns 2,3 showing the old and new labels).

    The A&K ordering of the levels is used in the various A&K tables above.

    The last table (8) (which uses the A&K ordering) addresses a question of the processes. The A&K labels refer to internal, mental ("cognitive") processes. There are corresponding external behavioural tasks, which are examples of A&K's "procedural knowledge type", and which you might require learners to do for instance in an assessment: shown in col.3. A longer definition of what I imagine these intellectual operations really mean is offered in the last column.

    What A&K felt they were doing

    A&K's contribution, in their own view at least, is:

    Thus A&K are fully aware that language is at best a partial clue to classifying objectives. In their main vignettes they often discuss this, and end by re-classifying an ILO from where they had first assigned it. Against this is a strong desire many of us feel to summarise every category by a single mnemonic word label -- specially useful when you are scanning a whole table trying to get an overview, to reflect on contrasts and coverage. This web page is primarily motivated by a desire to find or create such a useful visual overview, and it leads to a use of 1-word labels. This is however dangerous, problematic, and leads to many errors. A&K did NOT do this.


    For A&K, the knowledge type of metacognition is not the aspect of metacognition about self-regulation, not about managing one's learning; but is knowledge about the knowledge, about managing the application of each bit of knowledge. E.g. for arithmetic division, you can divide any two numbers except you mustn't divide by zero; the part-whole skill training issue that there is extra knowledge in how to connect the parts; near and far transfer issue is that there is extra knowledge in how to apply knowledge to each new case/context.

    They subdivide it into:

    1. Strategic: tag a procedure (both thinking and learning ones) with the metacognitive knowledge that it applies more widely.
    2. About "cognitive tasks": actually about test tasks. This is probably the "rules of the game" knowledge that students must pick up in order to do well in exams. What the tasks are that are being demanded (explicitly or implicitly); which tasks are relatively easy for this particular learner and her particular talents.
    3. Self-knowledge -- a person's estimate of their own: capability (self-efficacy); intrinsic interests; valuation of each type of task.

    Some diagrams


    Part B: Digital Bloom

    What about Bloom's taxonomy and current digital learning technology?

    One model for reviewing use of techno in teaching is the SAMR model (Substitution; Augmentation; Modification; Redefinition).

    Putting SAMR together with neo-Bloom you get Bloom's Digital Taxonomy:- Web page; or a 5-min video ...
    which first reviews the Bloom taxonomy; then the changes which the neo-Bloom revised version makes; and then how to link that to modern technology in education.

    Some further reading

    Part C: Criticisms of these taxonomies

    Bloom's taxonomy is about different types of learning objectives. It thus belongs within the part of education that believes teaching should be organised in this way, and that the goals of teaching should be explicit. (See here for counter arguments.)

    Alex Johnstone, Norman Reid

    For a different view, partly denying that Bloom's levels form a sequence (as opposed to an unordered set) see these notes by Norman Reid.

    Johnstone's categorisation of types of problem-solving subverts it in another way. Problem-solving is close to the heart of assessment tasks in science, but instead of classifying them as a developmental sequence with 6 levels, Johnstone suggested there were 8 types (only 2 of which were widely seen in teaching), made up of all combinations of 3 binary properties: whether the data were given or incomplete; whether the outcomes or goals of the task were given or "open" (decided by the learner before or during the task); and whether the methods were familiar or unfamiliar.

    Types of problem for solving (Johnstone 1993)
    Type Data Methods Outcomes/goals Skills bonus
    1 Given Familiar Given Recall of algorithms.
    2 Given Unfamiliar Given Looking for parallels to known methods.
    3 Incomplete Familiar Given Analysis of problem to decide what further data are required. Data seeking.
    4 Incomplete Unfamiliar Given Weighing up possible methods and then deciding on data required.
    5 Given Familiar Open Decision making about appropriate goals. Exploration of knowledge networks.
    6 Given Unfamiliar Open Decisions about goals and choices of appropriate methods. Exploration of knowledge and technique networks.
    7 Incomplete Familiar Open Once goals have been specified by the student these data are seen to be incomplete.
    8 Incomplete Unfamiliar Open Suggestion of goals and methods to get there; consequent need for additional data. All of the above skills.

    Johnstone,Alex. (1993) "Introduction" in Creative problem solving in chemistry: Solving problems through effective groupwork (London: Royal Society of Chemistry)

    What are Bloom taxonomies good for?

    They can help in writing:

    It is particularly useful for help in designing MCQs (multiple choice questions), since we tend to assume that MCQs can only test rote learning, but with care, you can test much "higher" kinds of learning.

    The helpfulness is generally by associating each Bloom category with keywords and using these to prompt a teacher's imagination. This operation works with teachers who apply masses of commonsense when using the prompts. However the inverse operation works very badly: scanning text for the keywords and automatically classifying with the Bloom level, because English isn't simplistic like that.

    In fact the "level" of a question actually depends on the processing it triggers in the learner's mind, and not on the words. Brain teasers (e.g. Mazur's "ConcepTests") have the surface form of simple factual questions, but trigger deep thinking about reasons for and against.

    Better alternatives

    Other prompting schemes also work, perhaps better.

    1. One good one is to take Laurillard's 12 learning activities, and use them as a checklist for examining whether a given course supports all of them. One big hole in Bloom's scheme which this exposes is the need to get courses to connect each topic with perceptual aspects of experience (see the section on this below).

    2. Another kind of checklist is of question types. One type which I've found very useful (as do audiences whom I've addressed) is the "assertion-reason question". Another type is the brain teaser I mentioned above.

    3. I want a different hierarchy, one based on critical thinking (Perry, Kuhn):
      1. Facts (and concepts)
      2. Reasons: that make a fact more or less likely to be true. Evidence.
      3. Knowledge of alternative, rival, possible conclusions
      4. Decision on which is more likely to be right, given all the evidence available.
      There is no trace in Bloom of [2], no admission that [3] is not only possible but important, and [4] is only discussed w.r.t. finding faults in conclusions, not to addressing the making of decisions under uncertainty.

    4. Steve Brindley used the following as a checklist for teaching (for lesson plans / learning designs).
      Steve Brindley's check list for learning designs
      Intellectual aims and goals Practical aims and goals Attitudes and interests
      Recall previous work Lab observations Individual / group work
      Problem recognition & awareness Record data Safety consciousness
      Problem solving Communication/ discussion Personal context
      Hypotheses, Prediction, Test Using powers of observation Practical problem solving.
      Generate ideas Using equipment Experimental learning
      Following practical instructions
      Recording facts / ideas in words
      Manipulating equipment

    Bloom level / question level / task level

    This section is unsatisfactory, and perhaps unnecessary. It would take a lot of work to properly articulate what is unsatisfactory from a modern perspective about Bloom's taxonomy. I'll just leave it in its present state.

    Bloom taxonomy levels don't really work; particularly not for classifying test questions.

    In fact the "level" of a question depends on the processing it triggers in the learner's mind, and not on the meanings of the words, let alone their surface form.

    In fact, especially in HE, each discipline typically has a core test activity type e.g. essay writing in History, problem solving using calculations in physics. These are usually a fairly good match to the Bloom level demanded, inherently requiring high level cognitive functioning; although the highest marks may require additional functioning that is not explicitly required by the question i.e. the student must know the implicit demands of the discipline (see below) e.g. whether to display originality or not, whether to redefine the exam question or whether this will be penalised.

    The fallacy of one word labels

    Most Bloom inspired work tries to identify single word labels for each taxonomic node or concept. This is fundamentally mistaken. If we suppose (and some may very reasonably deny this) that nevertheless there is a universal taxonomy to be discovered, it is a hidden abstract one referring to mental states of the learner which cannot be directly observed. The level of a question depends on the learner's mental processing and not on a question or task in isolation. This makes a taxonomy in practice much less useful than it seems because the operationalisation of a "level" cannot be translated into a task type or question format. Teachers with a deep if implicit grasp of this can be, and often are, usefully prompted by the taxonomy. Those without cannot be helped by it.

    Here's where good examples would fit, referring to the "level of a question".

    Not psychomotor but sensori-motor, and above all perception

    Bloom's attention to "psychomotor" while omitting perception seems to be a blunder. Each area of study has a personal, sensori-motor aspect (too often neglected). This is expressed in Laurillard's model as the private, experiential part of learning (as opposed to the public, abstract, conceptual one). However if you have to omit one domain, then all subjects have a perceptual part, but some like astronomy or geology do not have a motor one (a physically constructive part) and are entirely descriptive (you can't build a star in the lab, but you can observe them). I.e. perception is a more nearly universal aspect of learning than movement.

    For instance: In school biology (since Bloom's time) it has become compulsory for pupils to learn to distinguish in photomicrographs and diagrams eukaryote from prokaryote cells; animal from plant cells, etc. A physician must be able to recognise a disease in a patient, not just talk about it to other medics. A student of chemistry should know what gold looks like compared to copper. These things are taught and assessed. Bloom's taxonomy misses their important place, and A&K perpetuate this major omission.

    In using a checklist to review a course's content and assessment, I would always look at what connection it makes to students' personal experience, and especially to perceptual experience and skills.

    The above are educational arguments why splitting motor from perception is a blunder. It was an old-fashioned psychological tradition to split them; but one that is regarded as wrong by some important cognitive approaches since Bloom e.g. J.J.Gibson, and work on robotics. You can't build a practicable robot by splitting them: Rodney Brooks argued and demonstrated that you have to have layers each of which have both sensing and motor action, distinguished by how fast (but stupidly) they can respond.

    The psychomotor itself

    It seems to me that there is no trustworthy treatment of (psycho)motor knowledge. Dave (1970) is widely cited, following Bloom, whose student he was. However that work does not seem easily available. As my table 1 shows, there seems little agreement between Dave's stages and Harrow's.

    My impression is that there is no good work on this domain; and that defining a domain as motor without perception is a serious error both educationally and psychologically.

    Level of education and Bloom

    Some assert that Bloom levels map to levels of certificate in education systems. This must be wrong for reasons including these:


    The main list of Bloom levels expresses a set of values about what cognitive operations are more valuable than others. A&K have a slightly less rigid view of this, but still think there is a value ordering there. However:

    For instance, Arts schools (which at least in the UK award HE degrees), require for admission, in year 1, and in all exercises, that students exhibit different compositions from each other: i.e. they demand "creativity", the top Bloom level, pervasively. In contrast in psychology (for instance) students are drilled in an impersonal and conformist voice, not to put their own experiences forward as evidence, to produce the right answer not a creative one. Creativity is explicitly penalised.

    The old Bloom hierarchy put Evaluation at the top, above synthesis. That aligned with the view that the best paid jobs were executives, filtering what was offered by underlings to select the best things for the bosses' needs. As in the Renaissance, artists were relatively low paid employees. A more romantic view, expressed in the revised taxonomy, puts a higher value on creativity. Yet this fails to distinguish the synthesis a plumber shows (selecting a novel combination of off-the-shelf pipe and joint elements to suit a unique need) from the creativity described in product design texts, where design is not about assembling existing components, but finding a design that optimises multiple conflicting demands (that the product be cheap, be functional, be easily assembled, be easily cleaned, .....).

    In chemistry you will need to reproduce the symbol "W" for Tungsten, and paraphrasing that as "Wolfram" (which would demonstrate historical understanding) will be penalised: rote reproduction of the symbol is required. In many other disciplines however paraphrase will be a required demonstration. Not only does this show a difference in values between disciplines, but it would not be stated in test questions for chemistry: students must know this value.

    History as a discipline can usefully be considered as a training in writing "the History essay". Students will eventually be expected to take the question, redefine it (and state and justify their own redefinition), relate it to theories (probably to more than one) thus demonstrating "applying" and "analysing", and produce a weighted evalution. But they will get the highest marks by arguing a hypothesis of their own (demonstrating "creating") although it would never be called that. However in many other disciplines, this would be penalised as "not answering the question". This is not because Bloom is right and all academics across the world are wrong: it is because there are different values in different discipines, and Bloom's taxonomy does not and can not represent this.

    According to A&K, Bloom was aware that discipline area matters and said as much in writing. In that case, this criticism is not that they missed noticing the issue, but that their continued failure to address it does indeed undermine the whole enterprise.

    End summary

    What is good about Bloom's taxonomy is that it can serve as an effective prompt to teachers, to design better test questions and/or teaching.

    What is bad about it includes the points above, such as its neglect of perceptual learning, and its denial that different discipines require different kinds of thinking. And that even if the taxonomy is valid, it is a taxonomy of invisible aspects of learners' minds, and we cannot systematically and reliably translate this into question formats or other concrete teaching and learning activities.

    Its connection to improving teaching really amounts to a corrective for the pervasive tendency to design courses around "covering content", and instead prompts teachers to devote course time and effort to the mental tasks that are central to a given discipline: these are not merely higher types of cognitive processing than rote learning of answers, but more generally are what feedback and tutoring need to be concentrated on. (Content largely takes care of itself.) These are different in different disciplines, but what is common is the need to focus learning objectives, tests, and staff time on something more than reproduction of content.

    Appendix: Wanted -- killer examples of questions

    If I work further on this page, then the main thing is to generate good examples to match the plan list below, and thereby construct a demonstration of the fundamental wrongness of the Bloom taxonomic approach from a theoretical perspective (although it retains some pragmatic utility).

    Aims to fulfill

    Examples wanted

    1. A good assertion reason qu
    2. One with physics reasons but not on the course
    3. A qu. that could be physics OR other disc
    4. Antaki: types of why
    5. Aristotle: types of cause
    6. Health & Safety: types of cause
    7. wider: crystal healing, astrology
    8. Brain teaser: Mazur's ConcepTests (get real eg.s.)
    9. Brain teaser: Levis I.e. evoke strong reasoned attractions in 2 or more response items. I.e. the context of noticing internal contradiction
    10. Brain teaser: The 3 recent ones, like Levis (exp. growth, ...)
    11. (Howe) It's the context of peer challenge.


    Molesworth jokes here.

    Web site logical path: [] [~steve] [best] [this page]
    [Top of this page]