Recognition versus recall

At the start of Dr. Catherine Scott’s researchED Melbourne talk, she asked us all to draw a bicycle. I was sat next to John Sweller and neither of us had paper so we couldn’t join in, but the point was well made: all of us can recognise a bicycle but few of us can reproduce what it looks like. 

Scott developed a case that will be familiar to many readers. She displayed a diagram of a conventional model of the brain similar to the one that Dan Willingham uses in ‘Why Don’t Students Like School‘ except that she added a couple of arrows to highlight the process of forgetting. In this model, the brain takes in information from the environment that it processes in working memory. This is then exchanged with long term memory. The working memory is severely limited but the long-term memory is effectively limitless [In a pleasing coda, Sweller proposed a reason why the working memory is limited in his subsequent keynote].

A number of naysayers have criticised the model presented in Willingham’s book, stating that it lacks detail of sensory buffers, working memory structures and so on. Scott took us through these, some of which have instructional implications. Yet the simplified model is still a good one for teachers to come to terms with. 

Which brings us to an interesting question: how much is this a model and how much is it essentially the truth? In a recent blog post, Willingham explores the definition of learning. He tackles a number of candidates, including the idea from Paul Kirschner that learning is a change in long-term memory. I like this definition but Willingham criticises it on the basis that it is theory dependent; you first have to accept something like the working memory / long term memory model of the mind before it makes sense.

Yet this definition is far more powerful than those that depend upon features that are only sometimes associated with learning, such as the ability to transfer what has been learnt to other contexts. And it is the only definition that I am aware of that can accommodate both recall and recognition.

Scott’s talk was the first time that I had explicitly considered the distinction. Most of the time, we tend to define learning in terms of recall. And this makes sense: it’s no use being able to recognise that 7 x 7 = 49, you need to be able to retrieve that fact from long term memory at will if you want to reduce the cognitive burden of solving a maths problem.

However, recognition has its uses. We know that background knowledge is incredibly important for reading comprehension and this is why a number of schools are setting out to systematically build knowledge of the world along the lines of E. D. Hirsch’s core knowledge curriculum. Recognition of facts, connections and concepts is often going to be enough. 

And we may sometimes be testing for recognition when we think we are testing for recall. Scott suggested that multiple choice tests often do this. Fill-in-the-gaps assessments may well do this too.

Willingham has made the point that we frequently dismiss school learning on the basis that we think we’ve forgotten most of it. But this is grounded in a definition of learning that is just about recall. That knowledge may still be there, under the surface. We may be able to recognise it when we see it again and we may be able to convert it back to recall level learning much more quickly and easily than knowledge we have never learnt. 

Which is why I think that a definition of learning is not some esoteric point. Our folk theory of learning lets us down and teachers need something better.


44 thoughts on “Recognition versus recall

  1. In Special education we recoghnise a generic continuum of learning/teaching which goes from matching (put identical things together), to recognition (select the correct response from a group of possible responses) to production (produce the correct response).

  2. I think what I am about to write is merely an aside to this blog post, and it seems almost ridiculously obvious now that I am about to type it, but it seems to me our brains are wired to optimize recognition over recall.

    Or maybe it’s not so obvious, and maybe some in education have denigrated recognition as a cognitive task.

    I’m not an anthropologist, but I don’t picture ancient humans sitting around the fire, doing retrieval practice with types of poisonous plants or deadly predators. Instead, parents pointed at the actual things and said something along the lines of “good” or “stay away from that” and the offspring were expected to be able to recognize those things again and quickly be able to place them in the appropriate “good” or “stay away” categories.

    If the offspring got sufficiently proficient at this recognition they survived and had a much better chance of reproducing. So, we’ve survived as a species because of our ability to store information in long-term memory and then use that information to make inferences about our environment, including other agents in our environment.

    So, I would agree with your implication that any definition of learning must include recognition.

      1. Might be interesting to think about the tie-in here with Russell’s theory of descriptivism and Kripke’s arguments against. Do we define “things” in terms of their constituent parts (i.e. recall) or do we just “recognise” what a thing is?

  3. I always enjoy your blogs but think it is a shame that you so often end by implying that teachers are misguided and on the wrong track.

    1. Maybe Greg means they do not hear complete or good P.D. theory, ideas and information.
      Thousands of teachers do know this on an intuitive level and practise Greg’s ideas daily, but are often forced away from what they know to work.Although many teachers cannot articulate all these ideas, they certainly recognize them and say, “Finally, someone is saying what I have known all along.”

      1. Yes, but is seems an inherent assumption that Greg knows more than ‘teachers’. I think that it comes across as arrogant to generalise to such an extent. I thought I would give that feedback so he can address that if he chooses. There are plenty of teachers out there who are thinking, reading, intelligent, analytical and reflective professionals who may have taught for longer than he has and might be irritated at what could be considered to be a patronising tone. I am quite sure he is receptive to feedback. Writing is always a matter of audience. I suppose it depends who he considers his audience to be. I have met him and consider him to be a potential mover and shaker. I do hope he doesn’t catch the disease of ‘teachers know nothing’ but that he chooses to engage in the conversation. At this stage it may be 50/50.

      2. I have a similar critical style which is why I like this blog. I always assumed it came from my background in Physics. I know many find it annoying and direct and think it should be toned down but the issue then becomes a lack of clarity. Logical arguments are often quite curt which helps you focus on the premises built in and if you think they are reasonable or not.

        I work with many excellent colleagues but none have any real interest in research or evidence and will nearly always rationalize what they do as best practice without really digging underneath. Greg style is forthright and often blunt but it is scrupulously polite.

      3. Yes, but I am not objecting to his style, his content or his logic. I object to the assumption common to many of his threads that ALL teachers are ignorant of effective practice and suggest that phrases like ‘some teachers…’, and ‘there is a teaching element that…’ might come over better and would sound less self righteous.

      4. I understand. Could I ask you to look back for specific examples of this assumption. I believe you will struggle to find clear examples instead getting lots of sentences that imply it. You will likely think this implication is obvious, but it will not read like that to others. The point about style was to cover these kind of arguments and sentences that you have interpreted as arrogance but I interpreted as merely a logical argument. I hope this my point better. May I ask have you read up on the backfire effect it may be relevant to this discussion.

      5. Thank you, Michael. I just googled it. I think you are reading too much into my suggestion which was merely constructive criticism of what I saw as an over generalisation that I felt was less than inclusive and came across as slightly disrespectful. I imagine Greg will take it on as feedback from his readership.

      6. One interesting part of the effect was how it changed how we viewed idea based on how it matches our beliefs, in this regard but content and method of delivery is relavent. We have emotional responses powered by reactions that perceive the content as an actual immediate threat. There was a guide to combating the effect released which offered strategies for countering it but also explained the challenge s of arguing in general.

      7. I am not threatened by any of this content and I am a teacher and a teacher educator. Greg’s position is familiar which is why read his blog. It is also why I suggest he avoid generalisations that lump all teachers and teacher educators into the one basket of

      8. I don’t believe that I do overgeneralise in the way that you describe in this thread. I am sorry that you feel that I do. I will continue to write in the same way.

    2. Fiona,

      Maths results are falling all over the Western world. Why would you think that most teachers were on the right track?

      1. Fiona,
        You are asking a lot from Greg. He is a full time teacher, working towards a PhD, wrote a book on this, publishes one of the most detailed blogs on the topic. Yet you don’t want him to sound like he knows more than others.

        Unless someone putting in a similar amount of time you would expect Greg to be able to tell them something they don’t know.

        Greg is also clear where he lays the blame – he clearly believes teacher educators in general and some researchers are letting down everyone.

        If Greg is wrong in his ideas where are the compelling arguments from the teachers you refer to who may know more?

        If Greg is right why are these same teachers letting the problems Greg raises fester. Why don’t we hear from them?

        In Ontario the ideas such as PBL and inquiry discovery are the primary ideas used to describe the teaching approach that is promoted.

        The voices raising concerns are a few teachers and some uni profs in math departments.

      2. The empirical research shows that teacher education is not very good. Trainee teachers are not, on average, taught effective empirical based teaching strategies.

        As one academic told her trainee teachers:” Don’t worry about all that researched based teaching strategies, you will soon develop your own teaching style.”

        As an engineer, I could just imagine our professor saying that to trainee engineers: “Don’t worry about all that research about how to build bridges. You will soon learn your own building (design) style.”

        Another thing is that in engineering a pass mark is often something like 99.99%, where as in education 50% is often regarded as good enough.

        Just think if the 50% pass mark applied to building cars. If a production line building cars had half the cars with three wheels and half with five, this would be ok as the average is four.

        In schools, we have the situation where the top 20% of students are bored, and the bottom 20% continually fail in basic literacy and numeracy.

        If you want better teaching the teachers have to be taught what good empirical research has shown to be effective teaching strategies. Why then, are the schools of education afraid (?) of doing this?.

  4. Thanks for this Greg – it’s made me think about how I do my retrieval practice. I was moving towards MC questions as part of my retrieval practice, but actually, it’s more subtle than that: if I only need recognition (e.g. for general reading comprehension) MC is suitable, but if I want students to be able to recall and use knowledge (e.g. to solve a physics problem or write an essay on why Henry VIII split from Rome), then I need to use other quizzing/testing strategies. Is that your interpretation? Thanks, Ben

    1. There’s a danger here of an oversimplification. I think MC questions *could* be designed to test recall. But I think that, in teaching, we want to prove both recognition and recall so we should use a range of assessments. Clearly, a fact that was once available to recall may fade to recognition.

      1. you can also practice recognition to the point of recall. If I need my students to identify molecular substances correctly the best way is, as well as giving a definition, to practice looking at different diagrams and establishing which is molecular and which is not

  5. I wonder these are just shades of gray. Spontaneous recall without external stimulus would look like random thoughts, sometimes useful but not usually what you want. Recognition that can only recognize a specific bicycle is also not want we want in many cases.

    Perhaps someone has looked into some finely grained tests of stimulus to recall/recognition. The performance with increasingly general or vague stimulus to accurate recall would likely be something that looks at lot like expertise. An expert can recognize some useful common idea more reliably with less stimulus. (e.g. In physics this is a problem solved by principles of energy and momentum rather than force and acceleration.)

    1. Stan I was thinking about that while waiting for a game of football. Could we hypothesize that recall is just recognition but with vastly greater sensitivity. Internal thoughts could easily provide a trigger where no one else noticed one. Experts often bring unique insights to other fields, with varying success, thanks to this process. If nothing else refuting this would require an improvement in the definition of both recall and recognition.

      1. I think it is much more complex than a simple split between recall and recognition, no matter how useful drawing that distinction may be. For instance, we know that a lot of problem solving is about remembering similar problems that you have solved in the past. So recognition and recall are both involved in that process.

      2. In what way would you draw a clear distinction between recall and recognition. We were playing around with the idea that they are the same but with different levels of sensitivity to stimulus. If the idea is wrong then we need to identify the difference more clearly but if right talking about a spectrum may be more useful then using these two terms.

      3. I can imagine the distinction is similar to when I will be bald. I am not sure what will be the last non-bald hair count but I will know when I wasn’t and when I am.

      4. Only in the sense I am bald enough I may as well shave it all off. Complete baldness is rare and most people say they are bald when they have a unsightly patch. dosnt change my original point.

    2. My point was simply that while people may not agree on the exact boundary they will agree that some points on the continuum are one or the other. I follow S Pinker’s premise that when most people understand what a word means we don’t have to take the pedants too seriously and I am someone who generally enjoys a pedantic point.

      1. I recognized that but your analogy isn’t as clear as I think you intended it. If we are distinguishing between recall and recognition (premise: that is what this post is about) it is because it is useful to do so and leads to questions about what to focus on and how to layout delivery. If recall is simply hypersensitive recognition then we may be able to ignore the distinction and simply focus on improving recognition to a high level allowing recall to emerge. The argument that creativity naturally occurs out of a wide range of mastery is a similar idea. If a property naturally emerges from a prior skill is it necessary to focus on it.

        If this idea is wrong, and there is a distinct difference, then I was just hoping to define it so I could improve my thinking.

        There are lots of dichotomies or lenses we can use to analyse learning, we should investigate how useful this one is either by enhancing our understanding of the distinction or abandoning it, at least for the moment, as a useful tool.

        P.s if that hasn’t communicated my point try this. What benefit do I get by focusing on recognition and recall that I would not get simply by focusing on recall and improving its sensitivity to stimulus.

      2. The benefit is that you don’t assume that nothing has been learnt unless students can recall it. It’s worth reading the Willingham article that I link in the piece for a view of the problems caused by this.

      3. Recall under what conditions? In an exam, by being asked a technical question, by seeing a practiced problem. Maybe we make it harder, in a novel scenario where you need to identify similarities with practiced problems. These are all forms of recognition that lead to recall, but so is looking at a picture of a bicycle and recalling what it is useful for. The quote you supplied dosnt address this merely arguing for A high degree of recognition/recall. The more that you avoid the simple question I asked out the utility of the two words the more I am inclined to believe they are simply synonyms in this context.

      4. I believe I have answered your question but clearly I have not done so to your satisfaction. It is entirely your prerogative to draw whatever conclusions you wish.

  6. I think one thing we might be discussing is the difference between storage strength and retrieval strength. Robert Bjork claims that there’s evidence that everything we’ve learned remains in long-term memory, but that knowledge may become difficult or impossible to retrieve through disuse.

    Some examples:

    You probably have high storage and retrieval strength of your current address.

    If you meet someone at a party, you might have strong retrieval strength for her name that night, but since storage strength is low, you might forget her name the next day.

    For many of us, concern about how we’re coming across means that, when we meet someone new at a party, both storage and retrieval strength for her name are low. So the moment we walk away, we find ourselves asking, “What was her name again?”

    If you’re trying to remember who starred in that movie you loved years ago, you might have trouble pulling the actor’s name out of your brain, but as soon as your friend mentions the name you immediately know your friend is right. That’s high storage strength but low retrieval strength.

    If I understand what I’ve read on memory, retrieval practice will make both improve recall and make recognition easier and faster, but using recognition only to assess knowledge won’t necessarily improve retrieval strength.

    I don’t know about you folks, but this is why I want my doctor to have spent a lot of time on retrieval practice in medical school (and to have a lot of experience using the knowledge they committed to memory): when I go to her office with a baffling set of symptoms, I want her to have lots of knowledge about illnesses with both high storage and retrieval strength. That way she will very quickly recognize what those symptoms mean.

    1. Retrieval and storage seem more useful to me then recall and recognition. The distinction is more apperently and leads to necessary changes in how I teach.

  7. Apologies for my frustration in my last post Greg.

    My original idea was simply applying Occam’s razor. If we can use a continuum of recognition why bother with recall. While continuum’s can have labels at both ends they do not need them either. I have since changed my position as I now see we could use a continuum of recall instead. Unlike using low and high retrieval strength recall/recognition has no obvious affinity with either state.

    Willinghams quote can be re-written as nothing has been learned unless students can recognize and use something they have previous learned. (I realize that I have added extra text but this is actually implicit in the context of Willinghams quote. He’s talking about recalling learning without needing it excessively scaffolded for you, his original quote is also clearly more elegant then mine because of this).

    Your post seems to be offering recall/recognition as a useful dichotomy in additional to using low/high retrieval strength. The more I try to understand this, the less this seem so. We could define recall as high retrieval and recognition as low recall though really we should have four words to define all the combinations of retrieval and storage strength. I don’t believe anybody would find that particularly useful though. As an example of their reversibility I have re-wrote a key passage from this post.

    Most of the time, we tend to define learning in terms of recognition. And this makes sense: it’s no use being able to recall that 7 x 7 = 49 after seeing the answer, you need to be able to recognize a multiplication problem and retrieve the solution from long term memory at will if you want to solve 7 x 7.

    I don’t now think that it matters if we use retrieval, recall or recognition, as long as we add low or high strength to the phrase.

    1. Then we still disagree. There is an important phenomenon in education where a student would not be able to construct a correct response (recall) but would be able to recognise a correct response if presented with it (recognition). These may well map onto ideas of retrieval and storage strength but these latter two terms assume a particular cognitive science theory which many teachers will be unfamiliar with and might find confusing (I certainly have found them confusing). Regardless of this, teachers will be familiar with the concept I have described. The problem with assuming a position that nothing has been learnt unless correct responses can be constructed is that it implies that much teaching is ineffective because we often cannot recall specific concepts we learnt in school. This is not true because even if we can only recognise them they will still aid us with reading comprehension and we will likely be able to learn them back to recall level more easily if needed.

      1. Is the use of recall to describe constructing and recognition to describe remembering after prompting a established definition? If it has been used in experiments and analysis consistently then they now have technical definitions in this context and I will now use them as such. However if not then there is no reason to use one over the other. We can say students can recognise how to construct a solution and that they can recall after prompting. Any distinction would have to be spread, populisied and consistently used to be useful.

  8. Teachers will surely be able to understand low and high retrieval strength as easily as recall/recognition. We can even use low/high recall strength if you prefer. It is the low/high which makes it absolutely explicit rather then relying on a person understanding that in this context recall/recognition have specific meaning.

    CLT is more confusing because it is more detailed and makes more precise predictions. Replacing it with unclear generic terms that can be interpreted widely is not a virtue. If we remove storage strength and mention a simplified high/low retrieval/recall/recognition (your choice) people are hardly likely to be confused. (I do think this would be a waste though).

    My issue is only with the terminology you use as a lens for analysis. Your post did say that recall/recognition is not usually factored into definitions. After re-reading Kirshner’s article I found the following mentions of recall/recognition.

    In all of the many studies he reported, guided instruction not only produced more immediate recall of facts than unguided approaches,but also longer term transfer and problem-solving skills.

    Students learn to recognize which moves are required for particular problems, the basis for the acquisition of problem-solving schemas. (Both are on page 80)

    As you can see recall needs the adverb immediately before it and recognize is used in the same manner as recall. They are both simply synonyms in this context.

  9. In the 2014 edition of Visible learning page 116 Hattie gives precise definitions of both words (copied verbatim) in the context of education.

    recognise means to indicate that the material is known. This entails signalling, in a constant way, such as ticking a box on a multiple-choice test.
    But to recall means to produce, reconstruct, or rebuild.

    This is a technical definition that is far more precise then the generic definition and clears up my confusion. I suspect you will consider this obvious but it really isn’t and weakens the idea of a continuum.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )


Connecting to %s

This site uses Akismet to reduce spam. Learn how your comment data is processed.