Wednesday 6 January 2016

Tasks from the First Person Perspective (A Purple Peril)

The great snare of the psychologist is the confusion of his own standpoint with that of the mental fact about which he is making his report. I shall hereafter call this the ‘psychologist’s fallacy’ par excellence.
William James, The Principles of Psychology
This is a video of a baby trying bacon for the first time. The baby gets really really excited, and everyone around him goes 'Ha! Babies love bacon as much as the rest of us, this is great!'. And everyone laughs and cheers.

Except here's the thing. I think what this baby really likes is making his loved ones laugh and cheer. The bacon is fine, but I don't think it's the magical experience his parents assume. They are making the psychologist's fallacy: mistaking what you think is going on for what the person is actually experiencing. (It's our fallacy because our subject matter makes us uniquely susceptible.)

When people come into our labs to take part in experiments, we present them with a situation that we have designed to elicit a specific behaviour from them, and that we manipulate in various ways in order to probe the makeup of that behaviour. We therefore think we know what the person is doing: they are doing the thing we asked them to do. However, this isn't necessarily true, and in order to figure out what our participants did and why, we need to consider how they experienced the experiment. In effect, doing our science right means taking the first person perspective of our participants when we formulate our explanations. 

I take this idea primarily from Louise Barrett's excellent book, Beyond the Brain: How Body & Environment Shape Animal and Human Minds (which I reviewed here). The first couple of chapters spend a lot of time talking about anthropomorphism, and why it's a problem. To be honest, when I read the book I didn't quite know why Louise started with this. But over time, I've realised what an extraordinarily powerful point it is and we now talk about all the time. 

Peril Proposal: The psychologist's fallacy is real, but the ecological approach to understanding task dynamics and the information they create offers a useful framework for avoiding it while we science.

Task Analysis 
We talk a lot about tasks, and our 2013 paper made a big deal of the idea that our research has to start with a task analysis (Wilson & Golonka, 2013). One of the criticisms we've had back is that 'psychology already does this - I know what task my participants are engaged in'. But what people mean by this is 'I know what task I set my participants', and these are not necessarily the same thing. 

The core of the ecological approach is the notion of ecological information. Information variables are the higher order kinematic patterns in energy media such as light that are created by that energy interacting with the dynamics of objects and events in the area. Information is not identical to the dynamics (it can't be: it is a kinematic projection of dynamics) but the information is specific to those dynamics, and thus if you organise your behaviour with respect to the information as if it were the dynamics, it all works out pretty well. 

For us, a task analysis means 

  1. Define the task in dynamical terms (e.g. as I did in this paper on the affordances for throwing)
  2. Identifying the relevant dynamical properties that the organism needs to perceive in order to interact successfully with the task at hand
  3. Identify the information created by these dynamical properties
In Wilson & Golonka (2013), we used this to make the point that when you follow this path, you not only end up with productive science (information really does turn out to be where it's at) but you now have a principled way to identify any remaining explanatory gaps. You are dealing entirely in real parts and processes (to use our recent interest in mechanism talk) and the end result is more useful, more explanatory models. 

The Psychologist's Fallacy and Transfer of Learning
Gibson made psychology realise that the world creates useful information and that organisms use that information. In fact, from the first person perspective of the organism, information is all we have: there is no direct access to the dynamics of the world, there is only direct access to the kinematic projection of those dynamics. It works because that projection is governed by ecological laws and can specify the dynamics. But while we act as if we perceive the dynamics, we actually act because we detect the information and the details of our behaviour follow the information (see the development of the perception-action model of coordination, for an example of this). 

My current favourite example of this is the issue of transfer of learning. I am getting into this topic empirically (e.g. Snapp-Childs, Wilson & Bingham, 2015) and it's also coming up a lot in the context of applying the ecological approach to sports coaching and performance. I also just today listened to an episode of Rob Gray's excellent Perception-Action Podcast about transfer (that podcast is a must-listen, by the way!).

The issue at hand is, if I train you to do one thing, will that training translate to improved performance on related tasks? Everyone agrees that transfer will only happen if the two tasks overlap in some meaningful way but there is currently no agreement on what defines that meaningful way. 

This leads to a nice example of the kind of third person, fallacy-type thinking. People are often very surprised when skills don't transfer, between sports like tennis and badminton. They overlap in so many things, people say. The only real difference is that one involves a ball that moves one way, and another a shuttlecock that moves another. But everything else is basically the same, so it's weird skills don't transfer.

The error is to mistake what you think the tasks consist of with what the tasks clearly consist of as far as the player is concerned. These are clearly different things - you expect transfer, but none happens. The data are telling you your task analysis is incorrect, and you aren't noticing. 

From the first person perspective of the player, they detect and use information about the various dynamical events in the task. Information defines the form of the first person experience. This means that the relevant principle governing when transfer occurs should be do the two tasks overlap in the information they offer the player? This will happen to the extent the two tasks contain the same dynamical events. In tennis and badminton, this isn't the case: the dynamics of the flight of a tennis ball is entirely different from the dynamics of the flight of a shuttlecock and so too will the resulting information that can support intercepting it. I make this argument and present data to support it in Snapp-Childs, Wilson & Bingham, 2015 and have plans to develop the evidence base.

The proposal of this Peril is that the psychologist's fallacy is alive and well but can be successfully fought with an ecological task dynamics analysis who's goal is to identify the information a person is using to perform a behaviour. If you know the information, you can a) do science to see if people are using that information (e.g. a perturbation experiment) and b) be sure you are only including things in your explanations that are real parts and processes of the first person experience of your task. 


  1. This is a wonderful post, not least because it seems quite congenially consistent with my own interest in trying to figure out what people are doing by determining what perceptual variable(s) they are controlling. My approach to doing this is based on a control theory based methodology called the Test for the Controlled Variable (TCV). This method is illustrated in an on line demo at:

    Your approach seems to be based on your method of doing task analysis, which you describe as:

    1. Define the task in dynamical terms
    2. Identifying the relevant dynamical properties that the organism needs to perceive in order to interact successfully with the task at hand
    3. Identify the information created by these dynamical properties.

    I wonder if you could apply this process to the task we often use in our research; a compensatory tracking task like the one at I would like to see how your approach to determining the informational basis of a task compares to my approach to doing what seems to be essentially the same thing using the TCV.



    1. This task seems simple enough to try out (task dynamics is hard). I'll have a think when I get some time.

  2. That would be great! I really would like to know since I presume that your process of finding the informational basis of a task applies to all tasks, not just bimanual coordination.

    And I agree that task dynamics seems to be hard (it is for me, anyway) but I've found that just because something is "hard" doesn't mean it's right. Many people find control theory "hard" too but that's not what makes it right.