Keith Kendrick
Tel. (01223) 496385

• Contact via email


• Recent, selected Publications

Systems neuroscience: unlocking the secrets of neural encoding

A key obstacle to comprehending how the brain controls behavioural and physiological functions is that we still do not understand many of the basic principles whereby it represents and integrates information. Without this knowledge it is difficult to assess what causes the brain to malfunction following damage or pathology and this limits the potential for therapeutic advances. Our approach to providing this information is to combine methodologies for biological monitoring of the activities of significant components of neural networks with mathematical and computational approaches to help identify local and global encoding strategies used to represent information. We have focussed on how neural networks encode different odours since this is a relatively simple and well defined brain sensory system.

In order to record the electrical activity of neural networks in the olfactory system of rats which are responsible for identifying odours we have developed novel recording approaches using 30-48 microelectrodes assembled in a planar array with 250µM spacing (see Figure 1). This allows extracellular recordings to be made simultaneously from up to 120 output neurones (mitral cells) in a key olfactory processing region, the olfactory bulb, during odour processing in anaesthetised rats. We are also able to measure field potential activity at each electrode site giving us additional information about global patterns of pre- as well as post-synaptic activity across the whole of the electrode array.

To solve a major technical problem with adopting this biological measurement approach, namely automated analysis and sorting of information from different cells reliably in a huge and complex dataset, we have used a computational approach which combines a six feature principal component analysis with Kohonen clustering (see Figure 1). Analysis of frequency output of all the cells recorded by the arrays has shown that only small numbers (~10%) respond significantly to any given odour. The spatial patterns of groups of cells responding to each stimulus are different but overlap considerably. To analyse these patterns statistically we have developed a novel form of analysis of variance which can calculate a significance score associated with odour stimulation for individual and subsets of geometrically connected electrodes.

Figure 1 (Click to enlarge)

Each electrode in an array (A) detects spikes from the neurons close by; electrode 1 detects spikes from the blue neurons. In B & C, principle components and averaged waveforms are plotted for spikes sampled by one electrode and colour coded according to identified clusters. Artefacts in the data, denoted by the filled black circles, were not incorporated into clusters.

Does this mean that odour information is only encoded by altered firing frequencies of localised groups of cells within the network? An exciting outcome of our experimental approach has been the use of novel pattern recognition algorithms (Theme, Noldus) to establish that the majority of cells in the network increase the number and complexity of identifiable patterns in their output without altering their overall firing frequencies. This represents a large increase in information being encoded globally by the network as a result of odour stimulation. Power-spectrum analysis of the local field potentials recorded across the array has also confirmed that odour stimulation alters the contributions of key frequencies (theta and gamma) associated in many brain systems with learning and memory globally. This suggests the presence of extensive synchronised presynaptic activity throughout the network. Other global encoding principles that we have uncovered are currently being patented since they have potential application to improving design of biosensor and computer systems.

Understanding how these different global encoding principles are utilised by neural networks in the brain may ultimately help us explain why it can achieve such incredible amounts of information processing and perhaps lead to new approaches for identifying and treating dysfunction.

Olfactory and visual recognition memory

My group is using behavioural, electrophysiological, neurochemical and molecular approaches to study the brain systems, neuronal encoding strategies and transmitter and receptor signalling pathways (particularly glutamate receptors and nitric oxide) involved in olfactory and visual recognition memory. The main focus is to understand the precise nature of changes occurring in the neural networks processing sensory cues which allow recognition memory to form and then be recalled either just in the short-term (minutes or hours) or in the long-term (days, weeks or even years). How new memories integrate with existing ones, are influenced by emotional context and can combine with information from different sensory modalities are also being investigated.

Face recognition memory

The human brain has evolved specialised neural mechanisms for dealing with its most complex social tasks, namely communication (language) and individual recognition (faces). These special systems are of fundamental interest not only because their organisational principles may be different from those subserving other recognition functions, but also because they are of key importance for all aspects of our social interactions. With specialisation there tends to be an increased risk of malfunction and indeed problems with both face recognition (prosopagnosia) and language (aphasias) are not uncommon. To understand the detailed functions and vulnerabilities of such systems we need good animal models. While this may be a problem with language it is less so with face recognition.

Figure 2 (Click to enlarge)

Cells in the right temporal cortex encoding face categories or individual faces respond much faster and with greater synchrony than cells in the left
(*P<0.05 vs right hemisphere).

Although face recognition is a common feature in primates we are clearly limited in our ability, or justification, for using these species in large-scale studies to investigate its detailed function. We have now provided strong evidence that sheep possess very similar abilities to humans in using faces to recognise one another and also have the same specialised neural networks in their temporal and frontal cortises. Neural networks within these regions are dedicated to processing faces, as opposed to other visual objects, just as they are in monkeys and humans. The neural networks involved are organised hierarchically, with components classifying identity of specific categories of faces and individual faces taking longer to encode face stimuli than those simply involved in the process of distinguishing face from non-face stimuli.

These organisational principles in the brain face processing system endow it with remarkable speed in learning to recognise such complex stimuli and to remember large numbers of faces for long periods. Indeed, sheep can remember over 50 different sheep and 10 human faces for more than two years, and this is reflected in the fact that neural circuits encoding the faces of specific individuals continue to do so even when they have not been seen for over a year.

One of the key factors giving the brain face recognition system its advantages over other types of object recognition is that, like language, it involves one brain hemisphere more than the other. Thus sheep, like humans, pay attention more to visual cues from the right half of a familiar face (left visual field) for accurate recognition than the left half (right visual field). This is associated with a greater degree of activation in the face recognition area (within the temporal cortex) of the right hemisphere. Our electrophysiological experiments have shown that the speed and synchrony of activation changes in cells in the right hemisphere responding to faces of specific individual or face-categories is faster (by up to 400ms) and more temporally co-ordinated than in the left. This is not the case for cells responding generally to visual stimuli or generically to faces (see Figure). We are currently pursuing the hypothesis that individual recognition of faces is dependent on synchronised activity in the left hemisphere but that the right may be more important for processing the emotional and mnemonic consequences of recognition.


Recent, selected publications

Wu J, Kendrick KM, Feng J (2007) A novel approach to detect hot-spots in large-scale multivariate data.
BMC Bioinformatics 8:331
http://dx.doi.org/10.1186/1471-2105-8-331

Guevara-Guzman R, Arriaga V, Kendrick KM, Bernal C, Vega X, Mercado-Gomez OF, Rivas-Arancibia S (2009) Estradiol prevents ozone-induced increases in brain lipid peroxidation and impaired social recognition memory in female rats.
Neuroscience 159 940-950
http://dx.doi.org/10.1016/j.neuroscience.2009.01.047

Sanchez-Andrade G, Kendrick KM (2009) The main olfactory system and social learning in mammals.
Behavioural Brain Research 200 323-335
http://dx.doi.org/10.1016/j.bbr.2008.12.021

Ladroue C, Guo S, Kendrick KM, Feng J (2009) Beyond element-wise interactions: identifying complex interactions for biological processes.
PLoS One 4 e6899
http://dx.doi.org/10.1371/journal.pone.0006899

Ge T, Kendrick KM, Feng J (2009) A novel extended Granger Causal model approach demonstrates brain hemispheric differences during face recognition learning.
PLoS Computation Biology 5 e1000570
http://dx.doi.org/10.1371/journal.pcbi.1000570

Trabace L, Kendrick KM, Castrignano S, Colaianna M, De Giorgi A, Schiavone S, Lanni C, Cuomo V, Govoni S (2007) Soluble amyloid beta1-42 reduces dopamine levels in rat prefrontal cortex: relationship to nitric oxide.
Neuroscience 147 652-663
http://dx.doi.org/10.1016/j.neuroscience.2007.04.056

Brennan PA, Kendrick KM (2006) Mammalian social odours: attraction and individual recognition.
Philosophical Transactions of the Royal Society B: Biological Sciences 361 2061-2078
http://dx.doi.org/10.1098/rstb.2006.1931

Christen M, Nicol A, Kendrick KM, Ott T, Stoop R (2006) Odour encoding in olfactory neuronal networks beyond synchronization.
NeuroReport 17 1499-1502
http://www.neuroreport.com/pt/re/neuroreport/abstract.00001756-200610020-00008.htm

Tate AJ, Fischer H, Leigh AE, Kendrick KM (2006) Behavioural and neurophysiological evidence for face identity and face emotion processing in animals.
Philosophical Transactions of the Royal Society B: Biological Sciences 361 2155-2172
http://dx.doi.org/10.1098/rstb.2006.1937

Da Costa APC, Leigh AE, Man M-S, Kendrick KM (2004) Face pictures reduce behavioural, autonomic, endocrine and neural indices of stress and fear in sheep.
Proceedings of the Royal Society B: Biological Sciences 271 2077-2084
http://dx.doi.org/10.1098/rspb.2004.2831

Fabre-Nys C, Chesneau D, De La Riva C, Hinton MR, Locatelli A, Ohkura S, Kendrick KM (2003) Biphasic role of dopamine on female sexual behaviour via D2 receptors in the mediobasal hypothalamus.
Neuropharmacology 44 354-366
http://dx.doi.org/10.1016/S0028-3908(02)00410-0

Broad KD, Hinton MR, Keverne EB, Kendrick KM (2002) Involvement of the medial prefrontal cortex in mediating behavioural responses to odour cues rather than olfactory recognition memory.
Neuroscience 114 715-729
http://dx.doi.org/10.1016/S0306-4522(02)00231-2

Broad KD, Mimmack ML, Keverne EB, Kendrick KM (2002) Increased BDNF and trk-B mRNA expression in cortical and limbic regions following formation of a social recognition memory.
European Journal of Neuroscience 16 2166-2174
http://dx.doi.org/10.1046/j.1460-9568.2002.02311.x

Kendrick KM, Da Costa APC, Leigh AE, Hinton MR, Peirce JW (2001) Sheep don't forget a face.
Nature 414 165-166
http://dx.doi.org/10.1038/35102669

Trabace L, Kendrick KM (2000) Nitric oxide can differentially modulate striatal neurotransmitter concentrations via soluble guanylate cyclase and peroxynitrite formation.
Journal of Neurochemistry 75 1664-1674
http://dx.doi.org/ 10.1046/j.1471-4159.2000.0751664.x