The implications of a situated, embodied and dynamic perspective on cognition for modern neuroscience

Coherent and complex behaviour emerges from the mutual interaction of brain, body and
environment across multiple time-scales and not from within the brain alone [1, 2]. Consequently, the only genuine approach to understanding cognition is from a situated (interacting with an environment), embodied (having a body) and dynamic (spread through time), or SED, perspective [3, 4, 5]. These assertions have a long pedigree within philosophy [6, 7] and are beginning to have a major impact on the ideas and even the language of the cognitive sciences [8, 9]. Furthermore, the field of SED intelligence has grown immensely in the last decade and a fundamental conceptual framework described in terms of information and dynamical system theory has begun to take shape [10]. While the importance of extended dynamic processes and the body/environment context of an organism is at least implicitly understood in the neurosciences it has rarely been explicitly addressed outside of specific invertebrate systems, see [11] for a good review. A major goal of my work is to articulate the implications of this perspective for the neurosciences more generally.

To do this I start from a systems level rather than a system specific perspective more common
in traditional neuroethology (the study of the neural basis of behaviour) [11, 12]. However, I
subsequently attempt to ground all ideas within specific systems, draw quantitative correspondence between macroscopic dynamics and microscopic detail and identity the possible functional utility of the core ideas for neural computation. I provide two concrete examples of this approach below.

Cognition is spread through time

Typically experimental work within neuroscience involves the static presentation of stimuli to
passive, or even anaesthetised animals which has promoted the assumption that the brain is well approximated by steady-state measurements [13]. These ideas have been succinctly encapsulated by the attractor neural network framework where, following an input signal, the activation of a network of neurons settles to one of several discrete steady states from which the output is decoded [14, 15, 16]. However, we [17], and others [18, 5], have argued that systems that tend to settle into static states when examined in a static experimental setting may in fact be far from equilibrium when coupled to bodies and environments and the dynamics and function of a network may not be well represented by their static graph structure [19, 20]. Indeed there is increasing evidence that in many neural processes the dynamics away from steady state carry more computational/functional significance [21].

The workhorse for these ideas has been the olfactory system where odour identity is dynamic
and spread through time [18, 22, 23]. Emerging dynamical frameworks in olfaction have stressed the importance of transient dynamics away from baseline [24]. However, the impact of this work has been marginalised because the dynamical metaphor employed in this approach is drawn from ecology rather than neuroscience [24] and its is not clear, or at least not straightforward [25], to relate them to the micrcoscopic detail of neural substrates.

Recently we have developed an alternate dynamical systems framework for olfaction [26, 27].
We accomplished this by first developing a quantitative, multi-scale reduction of a biologically
detailed model of the olfactory system [26]. We found that experimentally observed dynamics are well described by stimulus-specific transient displacements of a single, globally stable fixed point [27, 28]. This framework is consistent with the putative mechanisms that encode odour identity suggested in the literature [29] and is invariant to both stimulus duration and stimulus intensity [30]. Furthermore the emphasis on the role of transient dynamics in computation resonates strongly with the reservoir computing paradigm developed in machine learning to classify time-series and dynamic data [31, 32]. We have an active collaboration with RMIC (Receptors and Membrane Ion Channels), Science Faculty at Angers University to test this framework in the moth olfactory system. We are also extending these ideas to account for odour categorisation in the olfactory bulb the dynamics of which, we argue, are not well accounted for by the attractor neural network framework [33].

Cognition arises from the interaction of brain, body and environment

The dynamic interaction of brain, body and environment serves as a powerful resource for coherent behaviour in addition to the brain [34, 17, 5]. Indeed, it has been claimed that coherent behaviour is not apriori dependent on the existence of internal dynamics within the brain at all. For example early work in behaviour based robotics by Rodney Brooks demonstrated that the integration of information from different sensory modalities can arise from the interaction of an agent with its environment indirectly [3]. More recent work has shown that behaviours that require memory can arise even when the nervous system lacks any explicit mechanism for state retention (for example a feed-forward system) given a rich enough environment [35]. More specifically we have shown that categorically discrete behaviours can arise even in the absence of corresponding discrete dynamical structures within the nervous system [17].

Studies in invertebrates have made considerable progress grounding these ideas in biological
systems [36, 37, 38] but they have had very little impact on vertebrate dominated mainstream
neuroscience. However one area of vertebrate neuroscience where the the interaction of brain and body environment is foregrounded is active perception [39]. Particularly, the whisker system of mice and rats has well circumscribed sensory/motor loops involving early sensory relays, thalamus and cortical regions [40]. The onset of whisking, and hence interactions between brain and body, is concomitant with qualitative changes in neural dynamics which constitute a so called brain state change [41]. All investigations of these phenomena so far have focussed on looking for internal triggers for these changes [42, 43]. However, we have argued that brain state changes are mediated by the onset of brain/body environment interactions (i.e. whisking) themselves [44, 45, 46]. This could comprise the first concrete example where the brain/body feedback is fundamental to the description of the neuronal phenomenon itself in a vertebrate system. We are currently collaborating with Fuji lab, RIKEN BSI (Brain Science Institute) utilising electro-cortiography (ECoG) to validate this theory by examining the impact of brain/machine/brain (BMBI) induced correlations on cortical brain dynamics.

Future Prospects

Recent technical innovations promise to give SED ideas renewed relevance for mainstream vertebrate neuroscience. The increasing prevalence of techniques that measure from neural populations with high temporal accuracy (multielectrode and optogenetic techniques) is bringing the neural dynamics of categorisation into sharp relief in other areas other than the olfactory system [47]. It is becoming clear that population responses are not well characterised by steady-state representations or by sets of discrete equilibria [48, 47]. I am extremely excited about extending the dynamical framework we develop in the olfactory systems to other sensory cortices.

The development of Brain Machine Interfaces (BMI), particularly for the control artificial prosthetics, involves real time decoding of ensembles of cortical neurons [49]. Visual feedback has been shown to be central to the stability of BMI interfaces [50, 51]. Furthermore the desire to provide real-time propriopceptive feedback through BMBI’s has placed dynamics of the sensorimotor loop at the centre of this research [50, 52].

Closed-loop experimental designs are becoming increasingly prominent within the neurosciences. Trackball set-ups allow virtual reality conditions for mice and have been adopted by several high profile labs [53, 54, 55]. Full body de-afferentation and environment simulation in the larval zebrafish promises whole brain measurement in the behaving animals [56] . Real-time feedback studies in songbird provide a partial but temporally precise insight into feedback dynamics [57]. Consequently, in vivo electrophysiology and optogenetics of behaving animals is quickly becoming an achievable gold standard in neuroscience. Preliminary studies already suggest that these methods challenge our pre-existing conceptions of neural function. Many neural phenomena have been shown to be contingent on the presence or absence of environmental feedback [57, 54]. We are now just beginning to bring to bear the theoretical framework we have developed for active perception to account for these differences.
In my opinion the ideas developed by taking an SED perspective on cognition are on the
ascendancy in modern neuroscience.


[1] J. J. Gibson. Observations on active touch. Psychological Review, 69(6):477–491, 1962.
[2] H. A. Simon. The Sciences of the Artificial. MIT Press, Cambridge, MA, 1969.
[3] R. A. Brooks. Intelligence without reason. In John Myopoulos and Ray Reiter, editors,
Proceedings of the 12th International Joint Conference on Artificial Intelligence (IJCAI-91),
pages 569–595, Sydney, Australia, 1991. Morgan Kaufmann publishers Inc.: San Mateo, CA, USA.
[4] R. D. Beer. Toward the evolution of dynamical neural networks for minimally cognitive behavior. In P. Maes, M. J. Mataric, J.-A. Meyer, J. B. Pollack, and S. W. Wilson, editors,
Fourth International Conference on Simulation of Adaptive Behavior, pages 421–429. MIT
Press, Cambridge, MA, 1996.
[5] R. D. Beer. The dynamics of active categorical perception in an evolved model agent (with
commentary and response). Adaptive Behavior, 4(11):209–243, 2003.
[6] M. Heidegger. Being and time. Harper Perennial Modern Classics, 2008.
[7] H. R. Maturana and F. J. Varela. Autopoiesis and Cognition The Realization of the Living.
Series: Boston Studies in the Philosophy of Science , Vol. 42, 1980.
[8] A. Clark. An embodied cognitive science?. Trends in cognitive sciences, 3(9):345–351, 1999.
[9] R. D. Beer. Dynamical approaches to cognitive science. Trends in Cognitive Sciences, 4(3):91–99, 2000.
[10] Conceptual and mathematical foundations of embodied intelligence. Workshop at the Max
Planck Institute, Leipzig., 2013.
[11] B. Webb. Can robots make good models of biological behaviour? Behavioural and Brain
Sciences, 24(6):1033–1050, 2001.
[12] J. D. Cliff. Computational neuroethology: A provisional manifesto. In J. A. Wilson and S. W.
Wislon, editors, From Animals to Animats: Proceedings of the First International Conference
on the Simulation of Adaptive Behaviour, pages 29–39. MIT Press, Cambridge, MA, 1990.
[13] M. Rabinovich, R. Huerta, and G. Laurent. Neuroscience – Transient dynamics for neural
processing. Science, 321(5885):48–50, JUL 4 2008.
[14] J. J. Hopfield. Neural networks and physical systems with emergent collective computational abilities. Proceedings of the National Academy of Sciences, USA, 79:2554–2558, 1982.
[15] J. J. Hopfield. Odor space and olfactory processing: Collective algorithms and neural implementation. Proceedings Of The National Academy Of Sciences Of The United States Of America, 96(22):12506–12511, OCT 26 1999.
[16] D. J. Amit. Modeling brain function: The world of attractor neural networks. Cambridge
University Press, 1992.
[17] C. L. Buckley, Peter Fine, S. Bullock, and E. Paolo. Monostable controllers for adaptive behaviour. In From Animals to Animats 10, volume 5040, pages 103–112. Springer Berlin Heidelberg, 2008.
[18] W. J. Freeman. Neurodynamics: An Exploration in Mesoscopic Brain Dynamics. London
Springer, 2000.
[19] L. Barnett, C. L. Buckley, and S. Bullock. Neural complexity and structural connectivity.
Phys. Rev. E, 79(5), 2009.
[20] L. Barnett, C. L. Buckley, and S. Bullock. Neural complexity: A graph theoretic interpretation. Phys. Rev. E, 83(4, Part 1), 2011.
[21] D. Durstewitz and G. Deco. Computational significance of transient dynamics in cortical
networks. European Journal of Neuroscience, 27(1):217–227, 2008.
[22] G Laurent and H Davidowitz. Encoding Of Olfactory Information With Oscillating Neural
Assemblies. Science, 265(5180):1872–1875, SEP 23 1994.
[23] G Laurent, M Stopfer, RW Friedrich, MI Rabinovich, A Volkovskii, and HDI Abarbanel. Odor
encoding as an active, dynamical process: Experiments, computation, and theory. Annual
Review of Neuroscience, 24:263–297, 2001.
[24] M. I. Rabinovich, A Volkovskii, P Lecanda, R Huerta, HDI Abarbanel, and G Laurent.
Dynamical encoding by networks of competing neuron groups: Winnerless competition. Phys.
Rev. Lett., 87(6), AUG 6 2001.
[25] T. Nowotny and M. I. Rabinovich. Dynamical origin of independent spiking and bursting
activity in neural microcircuits. Phys. Rev. Lett., 98(12), MAR 23 2007.
[26] C. L. Buckley and T Nowotny. Multiscale model of an inhibitory network shows optimal
properties near bifurcation. Physical Review Letters, 106, 238109, 2011.
[27] C. L. Buckley and T. Nowotny. Transient dynamics between displaced fixed points: An
alternate nonlinear dynamical framework for olfaction. Brain Research, 1434(SI):62–72, 2012.
[28] A. Zavada, C. L. Buckley, D. Martinez, J-P Rospars, and T. Nowotny. Competition-Based
Model of Pheromone Component Ratio Detection in the Moth. PLOS One, 6(2), 2011.
[29] O Mazor and G Laurent. Transient dynamics versus fixed points in odor representations by locust antennal lobe projection neurons. Neuron, 48(4):661–673, NOV 23 2005.
[30] M Stopfer, V Jayaraman, and G Laurent. Intensity versus identity coding in an olfactory
system. Neuron, 39(6):991–1004, SEP 11 2003.
[31] W. Maass, T. Natschlager, and H. Markram. Real-time computing without stable states:
A new framework for neural computation based on perturbations. Computation, 14:2531–60, 2002. 4
[32] H. Jaeger, M. Lukosevicius, D. Popovici, and U. Siewart. Optimization and application of
echo state networks with leaky-integrator neurons. Neural Networks, 20:335–352, 2007.
[33] J. Niessing and R. W. Friedrich. Olfactory pattern classification by discrete neuronal network states. Nature, 465(7294):47–53, May 2010.
[34] C. L. Buckley. A systemic analysis of the ideas imminent in neuromodulation. PhD thesis,
School of Electronics & Computer Science, University of Southampton, 2008.
[35] E. Izquierdo-Torres and E. Di Paolo. Is an embodied system ever purely reactive? In
M. Capcarrere, A. Freitas, P. J. Bentley, C. G. Johnson, and J. Timmis, editors, Eighth
European Conference on Artificial Life, pages 252–261. Springer, Heidelberg, 2005.
[36] V. A. Novakovic, G. P. Sutton, D. M. Neustadter, R. D. Beer, and H. J. Chiel. Mechanical
reconfiguration mediates swallowing and rejection in Aplysia californica. Journal Of Compara-
tive Physiology A- Neuroethology Sensory Neural And Behavioral Physiology, 192(8):857–870,
[37] F. J. Stewart, D. A. Baker, and B. Webb. A model of visual-olfactory integration for odour
localisation in free-flying fruit flies. Journal of Theoretical Biology, 213(11):1886–1900, 2010.
[38] Eduardo J. Izquierdo and Shawn R. Lockery. Evolution and Analysis of Minimal Neural
Circuits for Klinotaxis in Caenorhabditis elegans. Journal of Neuroscience, 30(39):12908–
12917, 2010.
[39] D. Kleinfeld, E. Ahissar, and M. E. Diamond. Active sensation: insights from the rodent
vibrissa sensorimotor system. Current Opinion In Neurobiology, 16(4):435–444, August 2006.
[40] Carl C. H. Petersen. The functional organization of the barrel cortex. Neuron, 56(2):339–355, 2007.
[41] S. Crochet and C. C. H. Petersen. Correlating whisker behavior with membrane potential in
barrel cortex of awake mice. Nature Neuroscience, 9(5):608–610, 2006.
[42] J. F. A. Poulet, L. M. J. Fernandez, S. Crochet, and C. C. H. Petersen. Thalamic control of
cortical states. Nature Neuroscience, 15(3):370–372, 2012.
[43] K. D. Harris and A. Thiele. Cortical state and attention. Nature Reviews Neuroscience,
12(9):509–523, SEP 2011.
[44] C. L. Buckley and T. Toyoizumi. The role of environmental feedback in a brain state switch
from passive to active sensing. in prep. In prep, 2013.
[45] C. L. Buckley. The role of environmental feedback in a brain state switch from passive to
active sensing. BARRELS XXV, SFN Satellite, New Orleans (2012).
[46] C. L. Buckley. The coherence of the brain and the environment, not the input statistics,
determines neural correlations. COSYNE, Salt Lake City (2012).
[47] K. D. Harris, P. Bartho, P. Chadderton, C. Curto, J. de la Rocha, L. Hollender, V. Itskov,
A. Luczak, S. L. Marguet, A. Renart, and S. Sakata. How do neurons work together? Lessons
from auditory cortex. Hearing Research, 271(1-2, Sp. Iss. SI):37–53, JAN 2011.
[48] L. Estebanez, S. El Boustani, A. Destexhe, and D. E. Shulz. Correlated input reveals coexisting coding schemes in a sensory cortex. Nature Neuroscience, 15(12):1691.
[49] M. A. Lebedev and M. A. L. Nicolelis. Brain-machine interfaces: past, present and future.
Trends In Neurosciences, 29(9):536–546, SEP 2006.
[50] M. A. L. Nicolelis and M. A. Lebedev. Principles of neural ensemble physiology underlying
the operation of brain-machine interfaces. Nature Reviews Neuroscience, 10(7):530–540, JUL
[51] M. D. Serruya, N. G. Hatsopoulos, L. Paninski, M.R. Fellows, and J.P. Donoghue. Instant
neural control of a movement signal. Nature, 416(6877):141–142, 2002.
[52] J. E. O’Doherty, M. A. Lebedev, P. J. Ifft, K. Z. Zhuang, S. Shokur, H. Bleuler, and
M. A. L. Nicolelis. Active tactile exploration using a brain-machine-brain interface. Nature, 479(7372):228–U106, NOV 10 2011.
[53] C. M. Niell and M. P. Stryker. Modulation of visual responses by behavioral state in mouse
visual cortex. Neuron, 65(4):472–479, February 2010.
[54] G. B. Keller, T. Bonhoeffer, and M. Hubener. Sensorimotor mismatch signals in primary
visual cortex of the behaving mouse. Neuron, 74(5):809–815, June 2012.
[55] C. D. Harvey, P. Coen, and D. W. Tank. Choice-specific sequences in parietal cortex during a virtual-navigation decision task. Nature, 484(7392):62–68, APR 5 2012.
[56] M. B. Ahrens, J. M. Li, M. B. Orger, D. N. Robson, A. F. Schier, F. Engert, and R. Portugues. Brain-wide neuronal dynamics during motor adaptation in zebrafish. Nature, 485(7399):471–480, MAY 24 2012.
[57] G. B. Keller and R. H. R. Hahnloser. Neural processing of auditory feedback during vocal
practice in a songbird. Nature, 457(7226):187–U78, JAN 8 2009.

%d bloggers like this: