The central role of expectation in cognition

I think we’re undergoing somewhat of a slow revolution in the cognitive sciences. The field is slowly coming to focus on the central role of prior expectation in cognition.

Evidence that prior expectation has a large effect on the interpretation of sensory input is by no means new, but it seems to me that people are focusing on prior expectation more and more when they theorize about the mind. For example:

* The recent post about hypnosis and the Stroop effect (the nytimes article focuses on this aspect of the work)
* The fact that most of the connections between lower-order sensory areas and higher-order areas are feedback connections from the higher to the lower
* The fact that imagining a visual image causes activation in (see [1], there’s a bunch of followup studies to that one too)
* This may be just a rumor, but I’ve heard that it’s been demonstrated that when you saccade, that near the end of the interim period when you cannot see, your primary visual cortical cells start firing in the patterns corresponding to what you expect to see at the new location. Does anyone have a source for that (I’ll ask the person who told me if they have a citation)?
* Jeff Hawkins’s effort to make theories of the mind centered on prediction
* The current fad in Bayesian analysis in theoretical cognitive science (which provides a mathematical framework for computing probabilities which take into account both prior expectation and evidence)

I don’t mean to imply that this potential paradigm shift is something that people are unaware of; indeed, Kosslyn, Hawkins and others have long been avid proponents of the view that sensory processing (as well as other aspects of cognition) is best understood as centered around prediction and prior expectation, not incoming sensory data.

[1] Kosslyn, S. M., Alpert, N. M., Thompson, W. L., Maljkovic, V., Weise, S. B., Chabris, C. F., Hamilton, S. E., Rauch, S. L., & Buonanno, F. S. (1993). Visual mental imagery activates topographically organized visual cortex: PET investigations. Journal of Cognitive Neuroscience, 5(3), 263–287.

8 thoughts on “The central role of expectation in cognition

  1. “Kosslyn, Hawkins and others have long been avid proponents of the view that sensory processing (as well as other aspects of cognition) is best understood as centered around prediction and prior expectation, not incoming sensory data.” — um, at least, that’s how i interpret their views, i could be wrong


  2. I was very interested to read your post. For my PhD I looked at the role of expectation in the control of smooth pursuit eye movements – these are the movements we use to follow a smoothly moving target such as a bird flying by. Normally we can’t perform these movements at will in the absence of a target – if we try we end up producing a series of jerky saccades instead. It seems we need the visual feedback from a smoothly moving target to make pursuit eye movements. BUT we can generate smooth pursuit movements in the absence of a target if we have a strong expectation that a target is going to appear. In this case we start moving our eyes smoothly several hundred milliseconds prior to the target, in anticipation. These anticipatory movements that are dependent on expectation can be preprogrammed in terms of speed, direction and timing based on both past experiences or on predictive cues. Both past experience and cues can inform our expectations thus allowing us to generate these anticipatory movements. References here:


  3. “The fact that most of the connections between lower-order sensory areas and higher-order areas are feedback connections from the higher to the lower”

    I’ve heard/read this from several sources without any actual citations of published evidence. I haven’t actually been able to find any such evidence in PubMed or Google Scholar searches. Can anyone give me some published examples or is this just “folk wisdom.” I’m not denying it’s possible, just that I can’t find the papers!


  4. hmm yeah i was looking for a ref for that too.

    this article cites some references; i haven’t followed and verified the citations yet, though:

    “Anatomical studies in several systems have shown that feedback projections from higher centers often vastly outnumber feedforward projections from the periphery (Cajal, 1909; Hollander, 1970; Ostapoff et al., 1990; Sherman and Guillary, 2002).”

    Cajal RS (1909) Histologie du système nerveux de l’Homme et des vertébrés. Paris: Maloine.

    Hollander H (1970) The projection from the visual cortex to the lateral geniculate body (LGB). An experimental study with silver impregnation methods in the cat. Exp Brain Res 10: 219-235.

    Ostapoff EM, Morest DK, Potashner SJ (1990) Uptake and retrograde transport of [3H]GABA from the cochlear nucleus to the superior olive in the guinea pig. J Chem Neuroanat 3: 285-295.

    Sherman SM, Guillary RW (2002) The role of the thalamus in the flow of information to the cortex. Philos Trans R Soc Lond B Biol Sci 357: 1695-1708.


  5. Starting from your refs I found some good support for control of the LGN by higher centers, and less satisfying info about V1 cortex. If one can’t distinguish local from long-range cortico-cortical synapses within cortex, then it’s hard to say how much input from higher centers there might be. But thinking further, considering synapse counts without knowing the relative strengths of the synapses could give a very biased picture, so maybe “most of the connections” isn’t an ideal basis for understanding function anyway. Hmmmm… needs further thought. Anyway, here’s where I went:

    From Sherman & Guillary (2002)…

    p. 1698: “Although geniculate relay cells are obviously in the business of relaying retinal input to the cortex, it is important for understanding thalamic relays (and possibly other relays) to appreciate that these retinal inputs, seen morphologically as RL terminals, provide only ca. 5–10% of all synaptic inputs to relay cells.” (ref. to: Van Horn, S. C., Eris¸ir, A. & Sherman, S. M. 2000 The relative distribution of synapses in the A-laminae of the lateral geniculate nucleus of the cat. J. Comp. Neurol. 416, 509–520. [This is a detailed study of the number of synapses and their sources. 83.7% of terminals are from the cortex or brainstem. Good stuff.])

    p. 1698: “Interestingly, for the cells in layer 4 of the striate cortex that are postsynaptic to the geniculate inputs, a similar pattern emerges. The driver inputs to these are the geniculate
    axons and yet these represent only 5–10% of the synaptic inputs to these cells in cats and monkeys.”(refs: Ahmed, B., Anderson, J.C., Douglas, R.J., Martin, K.A., Nelson, J.C. (1994). Polyneuronal innervation of spiny stellate neurons in cat visual cortex. J Comp Neurol 341(1):39–49. [only 6% of V1 layer 4 input is from the LGN, but they can’t distinguish between highly local cortico-cortical synapses and long-range ones from higher centers. Drat.] AND Latawiec, D., Martin, K.A., Meskenaite, V. (2000). Termination of the geniculocortical projection in the striate cortex of macaque monkey: A quantitative immunoelectron microscopic study. J Comp Neurol 419(3):306-19. [only 6.9-8.7% of V1 input is from LGN, and again no distinction between local and long-range cortical input]).

    More recent corticofugal auditory system reviews:
    Suga, N., Gao, E., Zhang, Y., Ma, X., Olsen, J.F. (2000). The corticofugal system for hearing: Recent progress. Proceedings of the National Academy of Sciences USA 97(22):11807–11814.
    He, J. (2003). Corticofugal modulation of the auditory thalamus. Experimental Brain Research 153(4):579–590.


  6. Thanks for checking those out Eric. Yeah, not being able to distingush inter- from intra- area certainly makes this statistic less meaningful for the top-down vs. bottom-up question. Makes you wonder if there are any other papers out there which do distinguish.


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