Walter J. Freeman's Neurodynamics and Neurophilosophy of Meaning

I've been interested in Walter Freeman's work for a while now, and having just read his book 'How brains make up their minds' (HBMUTM; note the double entendre) on holiday, I thought I'd have a go at summarizing some of his ideas. This is a pretty tall order, for at least four reasons: there are a lot of them, they are sometimes quite complicated, they are sometimes quite unconventional, and they are sometimes a bit loose. Here goes...

The scope of material covered in HBMUTM is very impressive - from basic physiology and anatomy to a psychology (of sorts) of perception and action via an extremely well-developed mathematical theory of neural population dynamics, with regular commentaries and contributions from from philosophy of science and philosophy of mind. However by the final chapter on 'knowledge and meaning in societies', Freeman's unbridled and speculative elaboration of his ideas does become quite tiresome. For me, the majority of interesting content can be found in chapters 2 ("Meaning and representation"), 3 ("Dynamics of neurons and neuron populations"), 4 ("Sensation and perception") and 5 ("Emotion and intentional action"), although the discussion of causality in chapter 6 ("Awareness, consciousness, and causality") is also interesting. I'm going to pick out and base the following on what I see as the two main strands to the book:

1. Freeman's general theory of brain function ('nonlinear neurodynamics') and of how meaningful percepts are formed from sensory stimuli
2. The running commentary contrasting three 'modes of intepreting experimental data from the neural and cognitive sciences about the nature of the mind': materialism, cognitivism, and pragmatism. 

Nonlinear neurodynamics and the neural basis of meaning




Most of Freeman's experimental work comes from intracranial EEG (also called 'ECoG' - 'electrocorticogram') recordings from olfactory cortex of rabbits when they detect and respond to conditioned stimuli (i.e. stimuli, normally chemical odourants, that they have been trained to associate with a reward). He considers that this association of a reward to the stimulus makes an otherwise irrelevant sensation (e.g. the smell of burnt wood) 'meaningful' to the animal. Behavioural saliency of this kind appears to be quite an important component of his notion of meaning; in fact at one point he says that the amplitude modulation patterns (see below) that characterize the meaning of an odourant to the animal actually disappear when that odourant is 'unconditioned' through a new training regime, or when it is fed to satiety, which remove the behavioural salience of the stimuli.

Whether one would actually need to condition stimuli in this way to invoke a salience-based response in humans is probably an open question. Perhaps simply following instructions in a visual object processing task would be enough to establish the salience of the stimuli; whereas for rabbits it is necessary just to get them to do the task? Clearly for people working on constructs labelled 'meaning' in other areas such as language or vision, it's very important that this terminological/experimental difference be borne in mind when reading Freeman's less anthropocentric use of the term. Irrespective of whether one agrees with every aspect of his theory of meaning in all its neuroscientific and philosophical intricacies (and I will not go into all of these here), however, I think most neuroscientists interested in brain systems at the meso- or macroscopic scale will find something interesting in this body of work.

For Freeman, meanings 'arise as a brain creates intentional behaviours and then changes itself in accordance with the sensory consequences of those behaviours' (pg. 9). 'Intentional' here is in the sense of 'intentionality' - not in the Brentanoan sense of the 'aboutness' of a representation (as generally used by contemporary philosophers), but rather in the original sense of Thomas Aquinas (who actually coined the term), as the 'directing of an action towards some future goal'. After an introductory chapter and an outline of his philosophical position in chapter 2, the core neuroscientific content of HBMUTM is given in chapters 3-5, over which Freeman gradually sets out what he describes as the '10 building blocks that allow us to understand how neural populations sustain the chaotic dynamics of intentionality'. These are generously summarized first at the end of chapter 2 (pgs. 37-38):

10 building blocks in the dynamics of intentionality:

1. The state transition of an excitatory population from a point attractor with zero activity to a non-zero point attractor with steady-state activity by positive feedback
2. The emergence of oscillation through negative feedbck between excitatory and inhibitory neural populations
3. The state transition from a point attractor to a limit cycle attractor that regulates steady-state oscillation of a mixed excitatory-inhibitory cortical population
4. The genesis of chaos as background activity by combined negative and positive feedback among three or more mixed excitatory-inhibitory populations
5. The distributed wave of chaotic dendritic activity that carries a spatial pattern of amplitude modulation made by the local heights of the wave
6. The increase in nonlinear feedback gain that is driven by input to a mixed population, which results in construction of an amplitude-modulation pattern as the first step in perception
7. The embodiment of meaning in amplitude modulation patterns of neural activity, which are shaped by synaptic interactions that have been modified through learning
8. Attenuation of microscopic sensory-driven activity and enhancement of macroscopic amplitude-modulation patterns by divergent-convergent cortical projections underlying solipsism
9. The divergence of corollary discharges in preafference followed by multisensory convergence into the entorhinal cortex as the basis for Gestalt formation
10. The formation of a sequence of global amplitude-modulation patterns of chaotic activity that integrates and directs the intentional state of an entire hemisphere

(bold in original text)

Those familiar with Freeman's other, more technically-oriented work, will notice that this summary contains some of the key components of his so-called 'K-sets' - a hierarchical framework for modelling the dynamics of mixed excitatory-inhibitory neural populations (see the scholarpedia article on K-sets for more on this). These provide the analytical basis for understanding the general phenomenological feature of the ECoG data that has really been the focus of Freeman's research career - what he calls (after Karl Lashley) mass action. This gave the title of his seminal 1975 book Mass Action in the Nervous System, but I recommend the scholarpedia article on mass action for a more up-to-date and concise summary. In short, 'mass action' refers to the observation  (originally in the olfactory cortex of rabbits, and now apparently in all sensory cortices) of the aperiodic emergence of global patterns covering large swathes of cortex. These global patterns are referred to as 'wave packets', where all points (ECoG sensors) show the same basic waveform (the 'carrier wave'), but with varying amplitudes (heights and troughs of the wave), resulting in so-called spatial 'amplitude modulation' (AM) patterns. It is these distributed spatial AM patterns that, for Freeman, constitute the 'meaning' of the stimulus to the animal;

"[Mass action] refers to the collective synaptic actions that neurons in the cortex exert on each other in vast numbers by synchronizing their firing of action potentials. In the aggregate, [mass action] is a powerful force that creates bursts of cortical neural activity that resemble the vortices of tornadoes and hurricanes. The bursts rapidly and repeatedly retrieve memories and bind them with sensory information into percepts. In this way, [mass action] expresses and transmits the meaning of sensory information in spatial patterns of cortical activity that resemble frames in a movie." (from the mass action scholarpedia article)

Above is an example of AM patterns from an 8x8 grid of electrodes spaced 0.5mm apart on the olfactory bulb. I got the figure from the scholarpedia article on mass action, but it is also in HBMUTM. On the left are example time series of the local field potentials, on the right are contour plots showing the mean spatial AM pattern across the grid. According to the figure legend,

'The pattern difference in Trial Set 1 (above to below) reflects the formation of a neural assembly during training. The pattern differences from Set 1 to Set 3 with the same control and CS [conditioned stimulus] inputs reflect permanent changes in memory with consolidation. The lack of AM pattern invariance with invariant stimuli shows that AM patterns are retrieved from memory and modified by input. They are not representations of sensory stimuli; they are memories that are created by the sensory cortex.' 

Thus, for Freeman, the spatial AM patterns are not to be understood as representations of the stimulus, since they are highly dependent on personal history and current goals, and are not invariant across trials. The fact that they appear to incorporate such a high degree of abstraction and contextual dependence is remarkable, given that (in olfactory cortex at least) they are only one synapse away from the sensory receptors.

I imagine a lot of people reading this will be asking themselves the same question I ask myself constantly, namely 'is this something observable in / relevant to EEG/MEG data / research in humans? The spatial AM patterns are (I believe) simply time series or induced power plots (left and right columns in the figure, respectively) of (gamma) band-pass filtered data. I think the classification with respect to the CS involves an ICA step (this isn't mentioned in HBMUTM). Clearly there is also some overlap with multivariate pattern analysis methods that are becoming popular in imaging.

The fact that the patterns are 'global', or at least cover large areas, implies some kind of connectivity between the different regions. There is a very revealing section in HBMUTM (pgs. 116-117) where Freeman likens his findings to those of a long list of scientists who have studied / are studying 'global processes underlying the unity of perception and action' - including Karl Pribram, Antonio Damasio, Jack Pettigrew, Paul Nunez, Stuart Hameroff & Roger Penrose, Dietrich Lehmann, Urs Ribary & Rodolfo Llinas, Catherine Tallon-Baudy, Mathias Muller, Francisco Varela, Steve Bressler, Moshe Abeles, and several others. So if like me you have an idea of the typical work attributed to these scientists, then you can get a bit more of an intuition as to what kind of phenomena Freeman thinks he is studying.

Interestingly, however, he does follow this up with the following comment:

"The formation of global AM patterns indicates that the sensorimotor and limbic areas of each hemisphere can rapidly enter into a cooperative state, which persists for perhaps a tenth of a second before dissolving to make way for the next state. This cooperation does not develop by entrainment of coupled oscillators into synchronous oscillations. as resonance is much too slow and the linear correlations between the waveforms from the different locations stay well above chance levels and do not fluctuate significantly. But it is not the level of correlation that changes with perception and action, but the global AM pattern, as the cooperation carries the entire hemisphere from one global chatoic attractor to the next. The genesis of macroscopic chaos by interacting brain parts does not require that they oscillate with identifical waveforms having linear correlations." (pg. 117)

My bolds and underlines this time. This caught my eye especially because my conception of global co-ordination amongst distributed brain systems absolutely does involve some combination of entrainment of coupled oscillators and resonance effects due to boundary conditions in contiguous synaptic action fields. I would say that is probably more or less the conventional view in the field at this point in time (anyone agree / / disagree here? ). Ok yes you can have synchrony without oscillations. But Freeman is suggesting something more radical here: abrupt and near-instantaneous transitions between globally coupled (and spatially amplitude-modulated) states; these are the 'frames' of sensation and perception. He sees this as no less than a paradigm shift, which perhaps makes it easier to appreciate why in more recent work he has been exploring different ways of modelling mass action borrowed from modern physics, such as percolation processes in cellular automata networks (with Robert Kozma), and (with Giuseppe Vitiello) as manifestations of spontaneous symmetry breaking. Deep.

Freeman's neurophilosophy and philosophy of neuroscience


According to the Stanford encyclopedia of philosophy:

"The literature distinguishes “philosophy of neuroscience” and “neurophilosophy.” The former concerns foundational issues within the neurosciences. The latter concerns application of neuroscientific concepts to traditional philosophical questions. Exploring various concepts of representation employed in neuroscientific theories is an example of the former. Examining implications of neurological syndromes for the concept of a unified self is an example of the latter."

In HBMUTM we see both of both of these. A number of intricate, albeit sometimes rather loose, positions are articulated on 'neurophilosophical' issues such as meaning, perception, consciousness, intentionality, and free will, amongst others. For me, though, the more interesting (but still rather eclectic) sentiments fall under the philosophy of neuroscience (/philosophy of psychology; these two often blur together) category. He states his agenda here early on:

"To relate the properties and operations of neurons and populations to the mental experience of meaning, I will contrast three modes of interpreting the experimental data we have from the neural and cognitive sciences about the nature of the mind. Within the general framework of the history of philosophy and psychology, there are three dominant views: materialist, cognitivist, and pragmatist, of which my theory falls into the pragmatist category." (pg. 23)

According to Freeman's scheme, materialists "view minds as physical flows, whether of matter, energy, of information, which have their sources in the world", cognitivists "argue that minds are made not of energy of matter, but of collections of representations that constitute symbols and images. They are software running on 'wetware'.", whereas pragmatists  "think that minds are dynamic structures that result from actions into the world." He attributes much of the pragmatist definition to Aquinas, and Aquinas' account of Aristotle's doctrine of 'active perception', "according to which the organism learns about the world and realizes its potential by its actions on the world." Also the similarities between the cognitivist and materialists are apparently due to the fact that  "over the past 50 years neurobiologists have borrowed  many of their interpretations of their data from cognitive scientists when they began to realize that the neuron doctrine wasn't working very well." 

This tripartite comparison forms a consistent thread that runs throughout the entire book, and is regularly returned to in the light of the particular content being discussed at various points. With respect to the core scientific topics of the book, spatial AM patterns in sensory cortices:

"...the data on the AM patterns can be interpreted in different ways, depending on your choice of philosophical premises."


According to the materialist view:


"these AM patterns reflect information processing....an odorant stimulus delivers information to the receptors, which process it by transducing it to action potentials. These pulses are transmitted to the bulb, where the information is bound into patterns and held, while it is being relayed by the tract to the cortex. The information stored in the cortex from previous stimuli is retreived and sent back to the bulb, where a comparison is made by correlation of the newly recieved AM pattern with each of a collection of retrieved AM patterns...The classification process is completed when the best match is found to identify an odourant. That best AM pattern is sent to other parts of the brain, where it serves to select and guide a fixed action pattern as a response to the stimulus."


for cognitivists:

"...each AM pattern represents an odourant. It is a symbol that signifies the presence of a source of food or danger. The receptor action potentials represent the features of the odourant, and the process by which the bulbar action potential are brought into synchrony through their synaptic interactions to represent an odour is feature binding. The integration of the features by a higher-order neuron makes it fire, and its activity represents that object that has the features.

...This approach cannot deal with the lack of invariance of AM patterns with respect to stimuli.


and for pragmatists:

"...the AM patterns are an early stage in the construction of meaning. They correspond to the 'affordances' advanced by J.J. Gibson in ecological psychology, by which an animal 'in-forms' itself as to what to do with or about an odourant, such as whether to eat the food or run from the predator giving the odorant information. They cannot be representations of odorants, because it is impossible to match them either with stimuli or with pulse pattens from receptor activation that convey stimuli to the cortex....They cannot be information, because that is discarded in the spatial integration performed by divergent-convergent pathways...They reveal the wings of attractors that are selected by the sensory pulses, each having a crater in the olfactory attractor landscape.
...In colloquial terms, the ingredients received by brains from their sensory cortices with which to make meanings are produce by the cortices. They are not direct transcriptions or impressions from the environment inside or outside the body. All that brains can known has been synthesized within themselves, in the form of hypotheses about the world and the outcomes of their own tests of the hypotheses..."

(pgs. 91-93) 

I personally find what Freeman is attempting to do here fascinating, but still can't help grinding my teeth a bit when reading it. I think the main reason for this reaction is that Freeman is (quite intentionally) mixing the history of philosophy (/ ideas) with current (/recent) debates and paradigms in psychology and neuroscience. In tracing the intellectual history of pragmatist thought, Freeman wants to thread an intellectual line from Aristotle through Aquinas to James, Gibson, Thelen, and himself (borrowing along the way from Dewey, Heidegger, Piaget, Merleau-Ponty). I think this is commendable and am not against such a project in principle, but unfortunately in practice what he ends up describing is times in danger of being somewhat incoherent - and, even worse, somewhat artificial. This is what I mean by mixing the history of ideas in philosophy with current debates in the sciences: yes there is a lively and thriving debate in psychology between cognitivists and dynamicists / enactivist / embodied cognition (any of which would be a more meaningful label than 'pragmatist' for the kind of ideas Freeman is describing), but I don't think it goes back to Aristotle or Aquinas. Similarly, I don't see the cognitivist/materialist distinction as really reflecting any current (or past, for that matter) debates in psychology or neuroscience, or the philosophy of psychology or neuroscience, that I'm aware of. I'm not sure how an 'information processing' perspective on brain function that doesn't invoke any concept of representations (as seems to be the case for Freeman's materialists) would work. And the perception-action cycle that for Freeman appears to go right to the heart of the pragmatist worldview ("Materialists and cognitivists view perception as a passive process that begins when a stimulus gives information that is transduced by receptors into a burst of neural activity, which cascades through the brainstem and thalamus into a sensory cortex...Pragmatists view perception as an active process, holding that humans and other animals maintain a stance of attention and expectation." (pg 101) has been right at the heart of cognitive psychology from very early one, as in e.g. the seminar work of Ulric Neisser in the 60s and 70s. I am sympathetic with some of Freeman's anti-representationalist sentiments, but I think the case is better articulated elsewhere (e.g. Chris Eliasmith, Tim van Gelder). 

Despite these gripes (make of them what you will), one thing that's certain is that it's always stimulating to hear people with strong opinions and different ideas. They probably bring out both the best and the worst in a writer/scientist/philosopher/person. I fear my opinions on Freeman's scientific and philosophical contributions may well fluctuate, as my own understanding and opinions develop themselves. One thing is certain though: they make you bloody well think.





Signing off,

john

Help, my brain won't normalise!

If you want to make strong, causal inferences about the brain networks supporting a cognitive function, you will probably need to recruit patients with brain lesions. Although fMRI indicates the networks correlated with a particular cognitive factor, fMRI in healthy controls cannot identify the regions that are essential. In contrast, if damage to a region impairs performance, this suggests a direct causal link between this region and cognition, and implies that the region is critical for preserved performance.

Nothing in life is easy, though, and analysing data from patients has its specific challenges. Patients are hard to recruit, and the size and location of their damage is unlikely to match conveniently to the brain areas you study. Once you have the data you need, your troubles are not over: this article deals with the challenge of normalising brain images that include large lesions.

Spatial normalisation is essential for group analysis of brain images. To detect effects in a given region, that region must be in the same place in every individual. In SPM, normalisation works by fitting each image to a template, first using a rough 12- parameter affine transformation (translation, rotation, stretching and skewing in three dimensions) and then using a more detailed non-linear warp (Friston et al. 2007). Warping applies local stretching and shrinking that minimizes the differences between the image and the template. This is great for healthy controls, but when the warping algorithm encounters a large, dark area in the brain it may try to shrink this area to make the image look "normal". Three main strategies have been proposed to deal with this:

Cost function masking

• Define a mask that covers voxels with damage
• Carry out warping ignoring voxels in the mask
• Warp calculated on intact voxels is extrapolated over the masked region
• Brett et al. (2001)

Unified normalisation

• Unified normalisation includes a segmentation step
• Warping is based on the segmented image: only considers tissue of interest
• This may effectively remove the lesion and make cost function masking redundant
• Ashburner & Friston (2005)

Warping regularisation

• Stricter regularisation penalises unlikely warps
• Makes shrinking of the lesion less likely

Crinion et al. (2007) made a rigorous comparison of unified and standard normalisation, at several levels of warping regulrisation, with or without cost function masking. Unified normalisation performed better than standard overall, reducing the variability in anatomic positioning and slightly improving fMRI results in a group analysis. Within unified normalisation, medium regularisation performed better than either low or high. Cost function masking had an effect only when regularisation was low, suggesting that medium or high regularisation is sufficient to prevent inappropriate warping of the image around the lesion.

The figure below shows images normalised with low (left), medium (middle) and high (right) warping regularisation. Shrinking of the lesion is obvious in low regulrisation.

Crinion et al. focussed on fMRI results as their end point, but warping may also influence VBM using structural images. Stricter warping regularisation may go too far and prevent fit of normal anatomic variations to the template. The figure below shows unified normalisation with strict (left) and medium (right) warping regularisation. The images were skull-stripped with the canonical SPM brain mask as part of preprocessing for VBM. This makes it obvious that strict regularisation prevented a good fit, such the dura was included in the skull-stripped image. [Images modified with photoshop to reduce identifiability].

Strict	Medium

Unlike fMRI data, where there is no signal of interest arising from the dura, VBM is intensity-based, and inclusion of dura may substantially alter results. The take-home message here is that you should check the normalisation of each individual image carefully, and consider the suitability of the normalisation method for the type of analysis to be run.

References

Friston KJ, Ashburner JT, Kiebel SJ, Nichols TE, Penny WD (2007) Statistical Parametric Mapping, First Edition. London, UK: Academic Press.

See also: chapter 3 in Human Brain Function, available online http://www.fil.ion.ucl.ac.uk/spm/doc/books/hbf2/

Brett M, Leff AP, Rorden C, Ashburner J (2001) Spatial normalization of brain images with focal lesions using cost function masking. Neuroimage 14:486-500.

Ashburner J, Friston KJ (2005) Unified segmentation. Neuroimage 26:839-851.

Crinion J, Ashbumer J, Leff A, Brett M, Price C, Friston K (2007) Spatial normalization of lesioned brains: Performance evaluation and impact on fMRI analyses. Neuroimage 37:866-875.

Perception and semantics in the ventral stream

One of the goals of cognitive neuroscience is to understand how we process visual objects, and what functional contributions different neural regions make. Our ability to perceive and interact with the world critically relies on the ventral processing stream through occipital and temporal cortices with increasingly anterior portions of the ventral stream responding to increasingly complex stimuli (Taylor, Moss & Tyler, 2007; Felleman & Van Essen, 1991; Tanaka, 1996; Tsunoda et al., 2001). As such, there is a relatively detailed account of how we process objects in a visual sense. However, an aspect of object recognition that is largely avoided concerns what an object means - it’s associated semantic knowledge. Instead, the dominant research strategy is to either focus on object recognition as a purely visual phenomenon, or study semantics without recourse to visual effects.

Recognising what a visual object is not only requires that objects are processed visually, but also that the semantic knowledge associated with the object is evoked. As such, a comprehensive account of how we recognise what an object is requires bringing together theories of visual object recognition, and cognitive models of semantic knowledge within the same neurobiological framework. This is the approach we’ve been developing (see Taylor, Moss & Tyler, 2007 for review) - understanding not only the cognitive contributions of different brain regions, but also how meaning emerges across time (e.g. Clarke, Taylor & Tyler, 2011). Uncovering how we understand what we see requires the development of a comprehensive systems-level account of how we get from perceiving an object, to understanding what it is, and requires the synthesis of cognitive theories and neurobiological models - a fundamental component of cognitive neuroscience.

Clarke, A., Taylor, K.I., & Tyler, L.K. (2011). The evolution of meaning: Spatiotemporal dynamics of visual object recognition Journal of Cognitive Neuroscience, 23(8), 1887-1899.

Felleman, D. J., & Van Essen, D. C. (1991). Distributed hierarchical processing in the primate cerebral cortex. Cerebral Cortex, 1, 1-47.

Tanaka, K. (1996). Inferotemporal cortex and object vision. Annual Review of Neuroscience, 19, 109-140.

Taylor, K. I., Moss, H. E., & Tyler, L. K. (2007). The conceptual structure account: a cognitive model of semantic memory and its neural instantiation. In J. Hart & M. Kraut (Eds.), The neural basis of semantic memory (pp. 265-301). Cambridge: Cambridge University Press.

Tsunoda, K., Yamane, Y., Nishizaki, M., & Tanifuji, M. (2001). Complex objects are represented in macaque inferotemporal cortex by the combination of feature columns. Nature Neuroscience, 4(8), 832-838.

Direct Electrical Stimulation: Insight into brain networks

I have just been going through a thought provoking new article by Hughes Duffau (2011), and thought I would provide a brief synopsis of the methods, and highlight a result or two. For those of you who are not very familiar with Direct Electrical Stimulation (DES) work, this would be one of the better places to start, as it is very well written and the reference section covers nearly the entire literature.

The first strength of the paper lies in the methods. Whereas the vast majority of human neuroimaging (MRI, EEG, MEG) attempts to measure cortical activity through the scalp, skull, and meninges, DES goes straight to the source: In patients undergoing brain surgery (awake patients to be specific), bipolar electrode tips deliver biphasic current pulses to the cortex and white matter of the human brain. While the current is being delivered, a patient’s performance on some task is being monitored. When the patient’s performance changes due to the stimulation, you have successfully mapped an “eloquent region.” In other words, that region’s neuronal activity plays a direct role in producing that behaviour.

Dr. Duffau's results using DES support a view of the brain that is very dynamic, and do not adhere to the localization hypothesis. To put it another way, complex human behaviours such as language rely on many different brain regions operating in concert. If one of the regions is damaged, it is possible to re-organize the circuit and preserve the function (as long as the white matter connections are left intact). Bear in mind that this is in relation to so-called frontal/executive functions (cognitive processing, language, emotion, etc), and basic sensory systems might operate far differently (not surprisingly, Dr. Duffau has published on this as well).

Below are a few references... I am including two other intriguing articles for those who want to know more about DES, and how the brain might be wired. As always, email me with any questions or just follow up with a comment.

Duffau H (2011) The “frontal syndrome” revisited: Lessons from electrostimulation mapping studies. Cortex (In Press).

Mandonnet E, Winler PA, Duffau H (2010) Direct electrical stimulation as an input gate into brain functional networks: principles, advantages, and limitations. Acta Neurochirurgica 152: 185-193.

Mandonnet E, Jbabdi S, Taillandier L, Galanaud D, Benali H, Capelle L, Duffau H (2006) Preoperative estimation of residual volume for WHO grade II glioma resected with intraoperative functional mapping. Neuro-Oncology 9: 63-69.

DTI preprocessing in FSL: B-vector correction

Using FSL to pre-process DTI data is easy to do, and the use of a standard FMRIB pipeline is ubiquitous in the literature. An ongoing concern with this common pre-processing pipeline is how to appropriately deal with eddy currents. FSL uses a 12 degree of freedom affine registration as a method of eddy current correction. By definition, therefore, it corrects for both eddy currents AND gross subject motion. So far this seems great... however...

It is really important to alter your b-vector file based on subject motion (Leemans and Jones 2009), but one should NEVER correct b-vectors based on eddy currents, since these vary across the brain (randomly) depending on a variety of factors. This can have profound effects on not only tractography, but also general diffusion modelling (DTIfit). Current wisdom on the FSL forum suggests that using the downloadable script “rotbvecs” will solve your b-vector correction issue. The problem is that this will correct for both subject motion AND eddy currents.

Here is a simple solution that will allow you to keep using FSL for preprocessing, and keep you up to date on appropriate b-vector corrections:

1) Create a new eddy_correct script that only uses 6-degrees of freedom, and call it eddy_correct_6dof. To do this, just add “-dof 6” to the following line of code:

${FSLDIR}/bin/flirt –in $i –ref ${output}_ref –nosearch –o $i –paddingsize 1

Then run the rotbvecs script as normal (this changes the bvecs file to reflect subject motion).

If you still want to correct for eddy_currents, simply run the regular eddy_correct script, and don’t touch the b-vector file. This might seem like a time consuming extra step, but probabilistic tractography is incredibly sensitive to the initial quality of the data. If you are looking for more detailed instructions, just send me a quick email (look to http://csl.psychol.cam.ac.uk/people/) for my contact info.

Of course ... one can also simply choose not to use FSL to pre-process the data... some great alternatives are emerging.

Variability and noise in the ageing brain

The vast majority of research reporting age-related changes to neural (e.g. BOLD activity, spike frequencies), behavioural (e.g. reaction times, error rates), or other biological variables is concerned with changes in measures of central tendency (e.g. means), that may change monotonically or nonlinearly with age. Changes to other statistical properties of these variables may also be relevant, however. Variability and/or (when definable) noise are two such properties, and whilst the study of these in the context of ageing is not new, recent findings have some interesting, slightly counterintuitive theoretical implications for how we understand the ageing process.


There are two main strands to this work that comes together nicely - one from cognitive psychology and cognitive neuroimaging (e.g. Garrett, McDonald, Grady), and the other from theoretical and computational neuroscience (e.g. Jirsa, Deco, Ghosh). Randy McIntosh seems to be the link between the two and his 201X review article ties up the two nicely.


The argument goes something like this:


Intra-individual neural variability increases with maturation and declines with senescence, and from a cognitive health point of view the decline is a 'bad' thing...


(Interjection: this is the counterintuitive part, since from both a cognitive and an engineering perspective we normally associate variability with imprecision, inaccuracy, and general lack of system 'integrity'. This leads to an important point:
exactly what variables and what kind of variability we're talking about here is an extremely important detail. More on this below.)


...Both the maturation and senescence patterns are interpreted in terms of there being an "optimal state characterized by a greater ability to transition between brain states and by increased dynamic range", whereas "older, slower, more behaviourally inconsistent participants lack a critical amount of brain variability for optimal neural function" (Garrett et al. 2011). The idea of there being an optimal level of variability/noise (possibly a stochastic resonance effect) has been developed by some of the contributing authors (McIntosh, Jirsa) in a number of simulation studies, where the brain is modelled as a network of coupled oscillators/dynamical systems that display neurophysiologically plausible behaviours such as intermittent synchrony. Their key soundbyte message is that possessing a small amount of endogeneously generated noise facilitates 'exploration of the brain's dynamic repertoire' - i.e. the space of possible brain states (see McIntosh et al. (2010) for a review that brings together the ageing data and the dynamical systems ideas most comprehensively). Other explanations offered by Garrett et al. (2011) for the reduced variability in old age are weakened functional connectivity (c.f. Fox et al. 2006; Nir et al. 2008), poorer neuronal signal detection (Li et al. 2006), or reduced dynamic range (Shew et al. 2009), and some vague comments linking it to 'Bayesian Brain' theories.


I think this is a fascinating and exciting hypothesis and am keen to see how it develops over the next few years. Here's my thoughts:




Comments + criticisms:




1. The list of empirical studies showing decreased variability/noise with (old) age is still quite low: 

• Garrett et al. (2010a):  examined effects of age on both mean- and standard deviation- (SD) based patterns of Blood Oxygenation Level Dependent (BOLD) signal in a simple visual fixation task. Although both the mean- and SD-based patterns were significantly predictive of age, the significance of the SD-based pattern was more than five times that of the mean-based pattern. Moreover, the brain regions comprising the multivariate spatial patterns identified by the SD-based and mean-based versions of their technique were non-overlapping - indicating that age-related changes in BOLD signal variability (measured here through standard deviation) constitute a distinct neural signature from changes in mean BOLD signal.

• Garrett et al. (2011): Similar to Garrett et al. (2010a), but different task and subjects (?)

• Park et al. (2010): Different group, different analysis methods: "Neural specificity was estimated based on how accurately multivariate pattern analysis identified neural activation patterns associated with specific experimental conditions.". Conclusion = loss of neural specificity (which they interpret as de-differentiation) contributes to reduced fluid processing ability in old age. 

• (for maturation: McIntosh et al (2008): found that intra-trial EEG signal variability decreased between the ages of 8 and 25 years. The level of variability was highly correlated with performance on a face memory task, and so the authors conclude that the move towards greater variability in brain activity during development improves cognitive flexibility and capacity)

• (in cardiac electrophysiology: Costa et al. (2005) develop a multi-scale entropy measure (which I believe is also used by Vakorin & McIntosh on EEG data) and apply it to cardiogram data, concluding that "The method consistently indicates a loss of complexity with aging, with an erratic cardiac arrhythmia satrial fibrillationd, and with a life-threatening syndrome scongestive heart failured....The results support a general “complexity-loss” theory of aging and disease")
• (in human motor control, Costa et al. (2007) apply the same measure to demonstrate that application of subsensory noise to the feet improves postural stability and multiscale complexity of sway fluctuations in healthy elderly subjects) 

2. Noise

One of the seminal recent works discussing variability and noise in the nervous system is Faisal et al. (2008), who define 'variability' as simply

“…changes in some measurable quantity, such as spike timing, or movement duration”. 


They also offer the following definition of noise (the subject of their review), following the Oxford English Dictionary:


 “…random or irregular fluctuations or disturbances which are not part of a signal […] or which interfere with or obscure a signal or more generally any distortions or additions which interfere with the transfer of information”.


Faisal et al. note that the term variability, unlike the term noise, is


a) agnostic to the source of the variability (random / deterministic), and


b) carries no implications of being either beneficial or detrimental.




We can add to these a further point, namely that 'variability', unlike 'noise'


c) does not necessitate a signal processing or information-theoretic interpretation of the phenomenon under investigation,


which is clearly not the case for the OED definition given above.


This final point reflects limitations of both the experimentalist and the theoretician, as it is simply not possible at our present level of understanding and technical finesse to define concretely the signals that a given pattern of neural activity might constitute or represent, beyond certain early sensory systems. And since you can only define 'noise' when you can define a 'signal', the fact that some of the papers discussed above use 'variability' and 'noise' are used interchangeably, is a bit sloppy.


The authors are of course aware of this, e.g. McIntosh et al. (2010) say


"An important clarification here is that while these three measures are sensitive to different aspects of  the brain signal, at present we take the liberty of considering them all as measures of brain noise. We acknowledge that this is not ideal, as noise has different implications at different levels of the system."

I think they sometimes they go for 'noise' rather than 'variability' because it sounds jazzier.

Another reason not to use the term noise is because of potential confusion with 'noisy brain' theories of sensory ageing, which go back several decades (e.g. Slathouse & Lichty 1985; Gregory?). Moreover, in electrophysiology studies of early sensory systems where signals can be straightforwardly defined, neurons older animals (monkeys) show increased spontaneous action potential firing rates, decreased stimulus selectivity, and lower signal-to-noise ratios (Schmolensky et al. 2000, Hua et al. 2006,Yu et al. 2006, Yang et al. 2009, Liang et al. 2010). Dopamine may also play a signal-enhancement role in cortical function, its steady decrease with age thus potentially also contributing to increased neural noise (e.g. Li et al.2001).




3. Relation to de-differentiation?


Park et al.  (2010) consider age-related increases in variability of multivariate spatial BOLD activity patterns as an index of 'decreased neural specificity', i.e. de-differentiation. Is this inconsistent with Garrett et al. 's ideas? This would seem to be the case for any spatial (rather than temporal) measure of variability.




References:

Costa et al. (2002a). Multiscale  entropy analysis of complex physiologic time series. Phys. Rev. Lett., 89: 068102

Costa et al. (2005). Multiscale entropy analysis of biological signals. Phys. Rev. E. Stat. Nonlin. Soft Matter Phys., 71: 021906, 2005.

Deco et al. (2009). Key role of coupling, delay, and noise in resting brain fluctuations. PNAS 106, 10302-10307

Faisal et al. (2008). Noise in the nervous system. Nat. Rev. Neurosci 9, 292-303

Garrett et al. (2010). Blood Oxygen Level-Dependent Signal Variability Is More than Just Noise. J. Neurosci. 30, 4914-4921.

Ghosh et al. (2008). Noise during Rest Enables the Exploration of the Brainʼs Dynamic Repertoire. PLoS Comp Biol 4(10), 12.

Grady et al. (2010). A Multivariate Analysis of Age-Related Differences in Default Mode and Task-Positive Networks across Multiple Cognitive Domains. Cereb. Cortex 20, 1432-144.

Li et al. (2006). A neurocomputational model of stochastic resonance and aging. Neurocomputing, 69(13-15), 1553-1560.

Liang et al. (2010). Aging affects the direction selectivity of MT cells in rhesus monkeys. Neurobiology of Aging 31, 863-873

MacDonald et al. (2003). Performance variability is related to change in cognition: Evidence from the Victoria Longitudinal Study. Psychology and Aging 18, 510-523.

McIntosh et al. (2008). Increased brain signal variability accompanies lower behavioral variability in development.  PLoS Comput. Biol., 4: e1000106.

McIntosh et al. (2010). The development of a noisy brain. Archives Italiennes De Biologie,148(3), 323-337.

Park et al. (2010). Neural specificity predicts fluid processing ability in older adults. Journal of Neuroscience, 30(27), 9253-9259.

Salthouse & & Lichty (1985). Tests of the neural noise hypothesis of age-related cognitive change. J Gerontol 40, 443-450.

  

Schmolesky et al. (2000). Degradation of stimulus selectivity of visual cortical cells in senescent rhesus monkeys. Nature Neuroscience 3, 384-390.

Yang et al. (2009). Aging Affects the Neural Representation of Speed in Macaque Area MT. Cereb. Cortex 19, 1957-1967

Yu et al. (2006). Functional degradation of extrastriate visual cortex in senescent rhesus monkeys. Neuroscience 140, 1023-1029.

See also:

Anstey et al. (1999). Sensorimotor Variables and Forced Expiratory Volume as Correlates of Speed, Accuracy, and Variability in Reaction Time Performance in Late Adulthood. Aging, Neuropsychology, and Cognition 6, 84.

Hultsch  & MacDonald (2004). Intraindividual variability in performance as a theoretical window onto cognitive aging. New Frontiers in Cognitive Aging 65–88

Nesselroade, & Salthouse (2004). Methodological and Theoretical Implications of Intraindividual Variability in Perceptual-Motor Performance. Journals of Gerontology - Series B Psychological Sciences and Social Sciences 59, (2004).

Rabbitt et al. (2001). There are stable individual differences in performance variability, both from moment to moment and from day to day. The Quarterly Journal of Experimental Psychology A 54, 981-1003 (2001).

Suckling et al. (2008). Endogenous multifractal brain dynamics are modulated by age, cholinergic blockade and cognitive performance. Journal of Neuroscience Methods 174, 292-300.


Vakorin et al. (2011). Variability of brain signals processed locally transforms into higher connectivity with brain development. J Neurosci. 31(17): 6405-13.

Welford, A.T (1984). Between bodily changes and performance: some possible reasons for slowing with age. Exp Aging Res 10, 73-88.

Welford, A.T. (1981). Signal, Noise, Performance, and Age. Human Factors: The Journal of the Human Factors and Ergonomics Society 23, 97-109

JG