Adrian Haith PhD

Adrian Haith


Research Interests

I am interested in computational modelling of human motor control and adaptation.  In particular, my PhD has focussed on how the motor system is able to adapt to different classes of disturbances - For instance, a kinematic disturbance like prism goggles, versus a dynamic disturbance such as a force field applied to the hand.

Other research interests of mine include optimal control theory, machine learning and computational neuroscience.

Publications:
2011
  Generative Probabilistic Modeling: Understanding Causal Sensorimotor Integration
Vijayakumar, S, Hospedales, T & Haith, A 2011, Generative Probabilistic Modeling: Understanding Causal Sensorimotor Integration. in J Trommershauser, K Kording & MS Landy (eds), Sensory Cue Integration. Oxford University Press, pp. 63-81.
This chapter argues that many aspects of human perception are best explained by adopting a modeling approach in which experimental subjects are assumed to possess a full generative probabilistic model of the task they are faced with, and that they use this model to make inferences about their environment and act optimally given the information available to them. It applies this generative modeling framework in two diverse settings—concurrent sensory and motor adaptation, and multisensory oddity detection—and shows, in both cases, that the data are best described by a full generative modeling approach.
General Information
Organisations: Neuroinformatics DTC.
Authors: Vijayakumar, S., Hospedales, Timothy & Haith, Adrian.
Number of pages: 19
Pages: 63-81
Publication Date: 2011
Publication Information
Category: Chapter
Original Language: English
2009
  A Theory of Impedance Control based on Internal Model Uncertainty
Mitrovic, D, Klanke, S, Vijayakumar, S & Haith, A 2009, A Theory of Impedance Control based on Internal Model Uncertainty. in ESF Intl. Workshop on Computational Principles of Sensorimotor Learning, Irsee, Germany.
General Information
Organisations: Institute of Perception, Action and Behaviour .
Authors: Mitrovic, D., Klanke, S., Vijayakumar, S. & Haith, Adrian.
Number of pages: 2
Publication Date: 2009
Publication Information
Category: Conference contribution
Original Language: English
  Implications of different classes of sensorimotor disturbance for cerebellar-based motor learning models
Haith, A & Vijayakumar, S 2009, 'Implications of different classes of sensorimotor disturbance for cerebellar-based motor learning models' Biological Cybernetics, vol 100, no. 1, pp. 81-95. DOI: 10.1007/s00422-008-0266-5
The exact role of the cerebellum in motor control and learning is not yet fully understood. The structure, connectivity and plasticity within cerebellar cortex has been extensively studied, but the patterns of connectivity and interaction with other brain structures, and the computational significance of these patterns, is less well known and a matter of debate. Two contrasting models of the role of the cerebellum in motor adaptation have previously been proposed. Most commonly, the cerebellum is employed in a purely feedforward pathway, with its output contributing directly to the outgoing motor command. The cerebellum must then learn an inverse model of the motor apparatus in order to achieve accurate control. More recently, Porrill et al. (Proc Biol Sci 271(1541):789-796, 2004) and Porrill et al. (PLoS Comput Biol 3:1935-1950, 2007a) and Porrill et al. (Neural Comput 19(1), 170-193, 2007b) have highlighted the potential importance of these recurrent connections by proposing an alternative architecture in which the cerebellum is embedded in a recurrent loop with brainstem control circuitry. In this framework, the feedforward connections are not necessary at all. The cerebellum must learn a forward model of the motor apparatus for accurate motor commands to be generated. We show here how these two models exhibit contrasting yet complimentary learning capabilities. Central to the differences in performance between architectures is that there are two distinct kinds of disturbance to which a motor system may need to adapt (1) changes in the relationship between the motor command and the observed outcome and (2) changes in the relationship between the stimulus and the desired outcome. The computational distinction between these two kinds of transformation is subtle and has therefore often been overlooked. However, the implications for learning turn out to be significant: learning with a feedforward architecture is robust following changes in the stimulus-desired outcome mapping but not necessarily the motor command-outcome mapping, while learning with a recurrent architecture is robust under changes in the motor command-outcome mapping but not necessarily the stimulus-desired outcome mapping. We first analyse these differences theoretically and through simulations in the vestibulo-ocular reflex (VOR), then illustrate how these same concepts apply more generally with a model of reaching movements.
General Information
Organisations: Institute of Perception, Action and Behaviour .
Authors: Haith, Adrian & Vijayakumar, Sethu.
Keywords: (Cerebellum, Motor adaptation, VOR, Kinematics, , , . )
Number of pages: 15
Pages: 81-95
Publication Date: Jan 2009
Publication Information
Category: Article
Journal: Biological Cybernetics
Volume: 100
Issue number: 1
ISSN: 0340-1200
Original Language: English
DOIs: 10.1007/s00422-008-0266-5
2008
  Integral equation methods for computing likelihoods and their derivatives in the stochastic integrate-and-fire model
Paninski, L, Haith, A & Szirtes, G 2008, 'Integral equation methods for computing likelihoods and their derivatives in the stochastic integrate-and-fire model' Journal of Computational Neuroscience, vol 24, no. 1, pp. 69-79. DOI: 10.1007/s10827-007-0042-x
We recently introduced likelihood-based methods for fitting stochastic integrate-and-fire models to spike train data. The key component of this method involves the likelihood that the model will emit a spike at a given time t. Computing this likelihood is equivalent to computing a Markov first passage time density (the probability that the model voltage crosses threshold for the first time at time t). Here we detail an improved method for computing this likelihood, based on solving a certain integral equation. This integral equation method has several advantages over the techniques discussed in our previous work: in particular, the new method has fewer free parameters and is easily differentiable (for gradient computations). The new method is also easily adaptable for the case in which the model conductance, not just the input current, is time-varying. Finally, we describe how to incorporate large deviations approximations to very small likelihoods.
General Information
Organisations: Neuroinformatics DTC.
Authors: Paninski, Liam, Haith, Adrian & Szirtes, Gabor.
Pages: 69-79
Publication Date: 1 Feb 2008
Publication Information
Category: Article
Journal: Journal of Computational Neuroscience
Volume: 24
Issue number: 1
ISSN: 0929-5313
Original Language: English
DOIs: 10.1007/s10827-007-0042-x
  A Bayesian Model of Multimodal Visuo-motor Adaptation
Haith, A & Vijayakumar, S 2008, A Bayesian Model of Multimodal Visuo-motor Adaptation. in Proc. 18th Meeting of the Society for Neural Control of Movement (NCM 2008). pp. 1.
We propose a model of multi-modal adaptation of reaching movements based on optimal Bayesian inference of the causes of errors. Our model accounts for the patterns of trial-to-trial adaptation as well as perceptual aftereffects in vision and proprioception when visual feedback is shifted or rotated.
General Information
Organisations: Institute of Perception, Action and Behaviour .
Authors: Haith, Adrian & Vijayakumar, Sethu.
Keywords: (Informatics, Computer Science. )
Number of pages: 1
Pages: 1
Publication Date: 2008
Publication Information
Category: Conference contribution
Original Language: English
  Unifying the Sensory and Motor Components of Sensorimotor Adaptation
Haith, A, Jackson, C, Miall, C & Vijayakumar, S 2008, Unifying the Sensory and Motor Components of Sensorimotor Adaptation. in Proc. Advances in Neural Information Processing Systems (NIPS '08).
Adaptation of visually guided reaching movements in novel visuomotor environments (e.g. wearing prism goggles) comprises not only motor adaptation but also substantial sensory adaptation, corresponding to shifts in the perceived spatial location of visual and proprioceptive cues. Previous computational models of the sensory component of visuomotor adaptation have assumed that it is driven purely by the discrepancy introduced between visual and proprioceptive estimates of hand position and is independent of any motor component of adaptation. We instead propose a unified model in which sensory and motor adaptation are jointly driven by optimal Bayesian estimation of the sensory and motor contributions to perceived errors. Our model is able to account for patterns of performance errors during visuomotor adaptation as well as the subsequent perceptual after-effects. This unified model also makes the surprising prediction that force field adaptation will elicit similar perceptual shifts, even though there is never any discrepancy between visual and proprioceptive observations. We confirm this prediction with an experiment.
General Information
Organisations: Institute of Perception, Action and Behaviour .
Authors: Haith, Adrian, Jackson, Carl, Miall, Chris & Vijayakumar, Sethu.
Number of pages: 8
Publication Date: 2008
Publication Information
Category: Conference contribution
Original Language: English
  Interactions between sensory and motor components of adaptation predicted by a Bayesian model
Haith, A, Jackson, C, Miall, C & Vijayakumar, S 2008, Interactions between sensory and motor components of adaptation predicted by a Bayesian model. in In: Workshop on Advances in Computational Motor Control (ACMC 2008).
General Information
Organisations: Institute of Perception, Action and Behaviour .
Authors: Haith, Adrian, Jackson, Carl, Miall, Chris & Vijayakumar, Sethu.
Keywords: (Informatics. )
Number of pages: 2
Publication Date: 2008
Publication Information
Category: Conference contribution
Original Language: English
2007
  Robustness of VOR and OKR adaptation under kinematics and dynamics transformations
Haith, A & Vijayakumar, S 2007, Robustness of VOR and OKR adaptation under kinematics and dynamics transformations. in Development and Learning, 2007. ICDL 2007. IEEE 6th International Conference on. pp. 37-42. DOI: 10.1109/DEVLRN.2007.4354055
Many computational models of vestibiilo-ocular reflex (VOR) adaptation have been proposed, however none of these models have explicitly highlighted the distinction between adaptation to dynamics transformations, in which the intrinsic properties of the oculomotor plant change, and kinematic transformations, in which the extrinsic relationship between head velocity and desired eye velocity changes (most VOR adaptation experiments use kinematic transformations to manipulate the desired response). We show that whether a transformation is kinematic or dynamic in nature has a strong impact upon the speed and stability of learning for different control architectures. Specifically, models based on a purely feedforward control architecture, as is commonly used in feedback-error learning (FEL), are guaranteed to be stable under kinematic transformations, but are susceptible to slow convergence and instability under dynamics transformations. On the other hand, models based on a recurrent cerebellar architecture [7] perform well under dynamics but not kinematics transformations. We apply this insight to derive a new model of the VOR/OKR system which is stable against transformations of both the plant dynamics and the task kinematics.
General Information
Organisations: Institute of Perception, Action and Behaviour .
Authors: Haith, Adrian & Vijayakumar, S..
Number of pages: 6
Pages: 37-42
Publication Date: 2007
Publication Information
Category: Conference contribution
Original Language: English
DOIs: 10.1109/DEVLRN.2007.4354055
2005
  Improved numberical methods for computing likelihoods in the stochastic integrate-and-fire model
Paninski, L, Haith, AM, Pillow, JW & Williams, CKI 2005, 'Improved numberical methods for computing likelihoods in the stochastic integrate-and-fire model' Computational and Systems Neuroscience (COSYNE) 2005, Salt Lake City, UT, United States, 17/03/05 - 22/03/05, .
A classic and recurring problem in theoretical neuroscience is to estimate the interspike interval (ISI) probability density: the probability that a white noise-driven integrate-and-fire-type neuronal model that has fired at time t=0 will not fire again until time t=T. This problem appears in a number of contexts, including firing rate computations, statistical model fitting, and decoding. In particular, Paninski et al. (Neural Comp. 2004) recently introduced likelihood-based methods for fitting stochastic integrate-and-fire models to spike train data; these techniques rely on the numerical computation of these ISI densities. Computing this likelihood is equivalent to computing a Markov first passage time density, the probability that the model voltage (a Markov process) crosses threshold for the first time at time t=T, given that the voltage was reset to some fixed subthreshold value at time t=0. Here we detail an improved numerical method for computing this likelihood, based on a technique of Plesser-Tanaka (Physics Letters A, 1997), and related to methods introduced by DiNardo, Ricciardi, and colleagues. We begin by noting that the ISI density uniquely solves a certain singular linear Volterra integral equation, then provide details on approximating this integral equation by a lower-triangular matrix equation, which may be solved efficiently on a computer. In addition, the gradient of this solution with respect to the model parameters may be computed efficiently via straightforward matrix perturbation techniques. This semi-analytic computation of the gradient greatly speeds numerical optimization of the model parameters in a maximum-likelihood setting and therefore enables consideration of models with many more parameters than has previously been feasible, e.g. in estimating spatio-temporal receptive fields of visual system neurons. This integral equation method has several advantages over the techniques discussed in our previous work (where we discussed two methods: one based on Gaussian integrals over "boxes" in a high-dimensional space, and the other on the numerical solution to a Fokker-Planck partial differential equation): the new method is more efficient, has fewer free parameters, and (as mentioned above) is much more easily differentiable. The new method is also easily adaptable for the case in which the model conductance, not just the input current, is allowed to vary as a function of time; this motivated us to develop a generalization of the main theorem in (Paninski et al, Neural Comp. 2004), that the likelihood has no non-global local maxima as a function of the model parameters, no matter what data were observed, even in the case of time-varying conductances. Finally, since the likelihood of a given spike train may be decomposed into a product over the likelihoods of each individual ISI, it is convenient to work with log-likelihoods. However, numerical errors in computing these small likelihoods can have a large deleterious effect on the overall likelihood computation in the log domain. Thus the computation of these very low-probability events must be handled carefully, both in the initialization stage of any maximization routine but also even near convergence to the maximum likelihood solution (since real data inevitably contains some outliers, when the neuron may have spiked at a highly unlikely time). In order to deal with this issue, we introduce a technique, based on the probabilistic theory of large deviations, which permits us to approximate these very small likelihoods on a logarithmic scale (for further details, see Paninski, this meeting). Once again, this large deviation approximation (along with its gradient) may be computed efficiently using simple linear-algebraic techniques.
General Information
Organisations: Institute for Adaptive and Neural Computation .
Authors: Paninski, L., Haith, Adrian M., Pillow, Jonathan W. & Williams, Christopher K. I..
Publication Date: 2005
Publication Information
Category: Poster
Original Language: English

Projects:
Modelling the role of the cerebellum in motor adaptation (PhD)