Professional Documents
Culture Documents
To cite this article: Lorne Whitehead & Saleh Banihani (2014) The evolution of
contralateral control of the body by the brain: Is it a protective mechanism?,
Laterality: Asymmetries of Body, Brain and Cognition, 19:3, 325-339, DOI:
10.1080/1357650X.2013.824461
Taylor & Francis makes every effort to ensure the accuracy of all the
information (the “Content”) contained in the publications on our platform.
However, Taylor & Francis, our agents, and our licensors make no
representations or warranties whatsoever as to the accuracy, completeness, or
suitability for any purpose of the Content. Any opinions and views expressed
in this publication are the opinions and views of the authors, and are not the
views of or endorsed by Taylor & Francis. The accuracy of the Content should
not be relied upon and should be independently verified with primary sources
of information. Taylor and Francis shall not be liable for any losses, actions,
claims, proceedings, demands, costs, expenses, damages, and other liabilities
whatsoever or howsoever caused arising directly or indirectly in connection
with, in relation to or arising out of the use of the Content.
This article may be used for research, teaching, and private study purposes.
Any substantial or systematic reproduction, redistribution, reselling, loan, sub-
licensing, systematic supply, or distribution in any form to anyone is expressly
forbidden. Terms & Conditions of access and use can be found at http://
www.tandfonline.com/page/terms-and-conditions
Downloaded by [Florida State University] at 10:18 24 December 2014
Laterality, 2014
Vol. 19, No. 3, 325339, http://dx.doi.org/10.1080/1357650X.2013.824461
Contralateral control, the arrangement whereby most of the human motor and
sensory fibres cross the midline in order to provide control for contralateral
portions of the body, presents a puzzle from an evolutionary perspective. What
caused such a counterintuitive and complex arrangement to become dominant? In
this paper we offer a new perspective on this question by showing that in a
complex interactive control system there could be a significant net survival
advantage with contralateral control, associated with the effect of injuries of
intermediate severity. In such cases an advantage could arise from a combination
of non-linear system response combined with correlations between injuries on the
same side of the head and body. We show that a simple mathematical model of
these ideas emulates such an advantage. Based on this model, we conclude that
effects of this kind are a plausible driving force for the evolution of contralateral
control.
the brain is to control the body via the nervous system, it is natural to consider
two possible modes of control, the most straightforward being ipsilateral
control in which left portions of the brain interact with left portions of the
body and similarly right with right. However, another possibility is
contralateral control in which left brain portions interact with right body
portions and vice versa.
Although both control types are possible, ipsilateral control might seem
to be a simpler, and possibly more likely, evolutionary development than
contralateral control. Furthermore, ipsilateral control could potentially be a
more adaptive arrangement because, in the event of a widespread injury on
Downloaded by [Florida State University] at 10:18 24 December 2014
just one side of the body, the non-damaged brain portion would be able to
control the corresponding non-damaged body portion on the other side. In
contrast, with contralateral control in such a situation, motor control of
both sides of the body would be impaired (on one side due to body damage
and on the other due to brain damage.) From this perspective it may seem
surprising that, over the course of evolution, contralateral control became
common in vertebrates and has come to dominate to the point that most of
the human motor and sensory fibres cross the midline in order to provide
control for contralateral portions of the body.
Motor control is provided by the reticulospinal, vestibulospinal, rubrosp-
inal, and corticospinal fibres. The vast majority of these fibres cross the
midline. The reticulospinal and vestibulospinal fibres control the trunk and
descend bilaterally. The mostly crossed rubrospinal fibres and the completely
crossed corticospinal fibres control the limbs. The bilateral reticulospinal
and vestibulospinal fibres are the most primitive and the oldest motor fibres
and are present even in the limbless species. The mostly crossed rubrospinal
fibres are present only in species with pseudolimbs or limbs and occur after
the reticulospinal and vestibulospinal fibres and before the corticospinal
fibres (Serge, Raineteau & Jabaudon, 2005). This may indicate that the
evolution of the rubrospinal fibres coincides with the evolution of
pseudolimbs, as are found, for example, in some segmented worms. The
completely crossed corticospinal fibres are the most recent motor fibres in
evolution and are present only in species with limbs.
This coincidence between evolution of more crossed fibres and evolution
of pseudolimbs and limbs represents the basis for a recent hypothesis
(Banihani, 2009) to explain the phenomena of fibre crossings in the central
nervous system (CNS). According to this hypothesis the occurrence of
separated parts of the body such as limbs and eyes created a need for
functional binding between the two halves of the CNS, and in some way
this need served as an evolutionary stimulus leading to the crossed
arrangement for the sensory and motor fibres that serve such separated
parts. This hypothesis is consistent with the fact that all fibres that carry
separated information (such as spinothalamic fibres, medial lemniscus,
EVOLUTION OF CONTRALATERAL CONTROL 327
fibres that carry monocular vision, and corticospinal fibres) do cross the
midline and all fibres that carry shared information (such as olfactory fibres
and fibres that carry binocular vision) do not. However it is not clear how
the need for functional coordination and information transfer between the
two halves of the CNS would lead to the evolution of the contralateral
control arrangement.
Other hypotheses that explain the emergence of contralateral control
concentrate mainly on crossing of sensory pathways. These are often called
the ‘‘avoidance behaviour hypothesis’’ (Cajal, 1899), the ‘‘image-forming eye
hypothesis’’ (Sarnat & Nesky, 1976), the ‘‘binocularity (binocular vision)
Downloaded by [Florida State University] at 10:18 24 December 2014
survival. Mobility also enables an animal to find food, water, and suitable
mates. For all these reasons there would be a significant evolutionary
advantage favouring the development of anatomical features that lessen the
chance of losing mobility. Here we propose that contralateral control could
provide such an advantage. Our argument relies on general considerations of
the mobility task, from the perspective of control theory.
Animal mobility is made possible by a complex system of interactions
within and between the body and the brain, one that involves control
information passing from the brain to the body and sensory information
passing from the body to the brain. The incredible accuracy required for
successful rapid navigation in a complex environment is achieved by
countless intertwined feedback loops. In general, such complex systems
have certain characteristics that arise from their use of the feedback
principle as well as from various forms of redundancy. Such systems can
be remarkably resilient and also remarkably adaptive to changes in
circumstances. In the very common example of operational amplifiers, for
example, the gain is normally determined almost entirely by a negative
feedback loop, so in the event that a temperature increase changes the
intrinsic gain of the operational amplifier itself, the overall circuit
performance remains essentially unchanged. In the case of animal mobility
there would have been strong selection pressure towards such characteristics.
For example, if muscles are fatigued and therefore less responsive than
usual, the brain, through feedback, will compensate by appropriately
adjusting its control signals. If a limb is slightly injured, the brain may
select a gait that reduces risk of further injury, favouring the hurt limb until
it can recover. The well-known phenomenon of plasticity in both brain and
muscle tissue contributes to such compensation. Thus complex systems
involving feedback, adjustment, compensation, and redundancy are needed
to support robust mobility.
There are two overarching considerations concerning the performance of
such a system in the event of injury. The first involves the diverse range of
possible injury severity. From an evolutionary perspective, injuries that are
EVOLUTION OF CONTRALATERAL CONTROL 329
severe enough to prevent mobility are irrelevant, because they are effectively
fatal. Similarly, injuries that are too minor to noticeably impair mobility are
also irrelevant. Here we must consider injuries that are of an intermediate
nature; those that are serious enough to reduce mobility but only to a
limited extent, so that the injured animal may be able to survive long
enough to heal and regain full function. The second consideration is that
the brainbody system has considerable reserve capacity, which means that
it is possible for moderate injuries to cause little degradation in mobility,
and thus survival and recovery may indeed be possible. In summary, we
need to consider the effect of injuries of intermediate severity on a complex
Downloaded by [Florida State University] at 10:18 24 December 2014
interactive system of brain and body, one that has significant reserve
capacity.
Often it is helpful to try to model such dependencies in an approximate
quantitative manner. This is not to suggest that such models are literal
representations of actual characteristics, but nevertheless they may be
helpful in showing certain simple and important regularities. For example,
the field of control theory contains significant mathematical analysis of the
advantages and subtleties of complex feedback-based control systems
(Åström & Murray, 2008). Generally, a key theme of such work is that
continual adjustment, based on multiple redundant forms of feedback and
over multiple time frames, makes it possible for intrinsically inaccurate
system components to work together to produce very accurate, reliable
outcomes. An animal running rapidly through the jungle is an extremely
sophisticated example of such a system.
Such systems typically have another key characteristic that is essential for
their robustness: a non-linear response to intensity of injury. This is perhaps
best illustrated in the findings in the field of percolation theory (Kirkpatrick,
1973), in which a model network is studied. For example, the model could be
a lattice of connected resistors in which a certain fraction, f, of the resistors is
randomly removed. The electrical resistance of the overall network is then
studied as a function of f. There are many different variations, most of which
show a similar behaviour: When a small fraction of the resistors is destroyed,
there is only a small change in resistance, but as the fraction f approaches
some critical value the resistance increases catastrophically*the system’s
ability to operate is lost. This type of non-linearity is a common
characteristic of complex, redundant systems. From this perspective, in
order to specifically consider the case of the brainbody control problem, it
is helpful to consider a simple mathematical model of how the degree of
mobility impairment may be related to the severity of injury. In doing so we
are not attempting to prove that any one simple model in fact predominated
throughout evolution*we simply wish to demonstrate that the plausibility
of models that favour contralateral control.
330 WHITEHEAD AND BANIHANI
W ¼ 1 fn (1)
In this model n is a fixed number, greater than 1, and its size indicates the
degree of non-linearity of the system response. Recall that f is a number
between 0 and 1 and so, therefore, is W. We will also assume that, after an
injury, an animal can survive and procreate only if W is greater than a
minimum required value of Ws. Figure 1 shows the relationship of W, f, and
Ws for an example value of n 4 and Ws 0.8.
To complete this simple picture, let us now assume that, when an injury
occurs, the amount of damage, f, is determined by chance, with an equal
likelihood of any value occurring within its range from 0 and 1. This means
Figure 1. Graph of wellness fraction W vs injury fraction f for the case n 4, with the straight line
segment indicating the value Ws.
EVOLUTION OF CONTRALATERAL CONTROL 331
that the probability of the value of f being below a given value x (chosen
between 0 and 1), is simply that value x. (For example, there is a 20% chance
that the randomly determined value of f will be less than 20%.) In turn,
considering equation (1) and the stated requirement for survival that, after
injury, W must exceed Ws, it then follows that the probability of survival, Ps,
will be given by:
p ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
Ps ¼ n
1 Ws (2)
Downloaded by [Florida State University] at 10:18 24 December 2014
calculation that takes into account the degree to which the random values of
fbrain and fbody are correlated.
It is simplest to first consider the case in which of fbrain and fbody are
perfectly correlated; in other words they are always equal to the same value f.
In this case it is easy to show that equation (3) is equivalent to equation (1)
and so the probability of survival is the same as that discussed above*so if
we again take the example value for n of 4, and if we choose 0.8 for Ws, then
Ps $ 0×67.
Let us now consider, from the perspective of correlation, the opposite
extreme in which there is no correlation whatsoever between the severity of
Downloaded by [Florida State University] at 10:18 24 December 2014
injury of the two areas. In this case the calculation is a bit more complex (see
Appendix A), but the results can still be expressed in full form. The summary
for the two cases is as follows:
p ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
ðaÞ full correlation : Ps ¼ n
1 Ws (4a)
2=n
ðbÞ no correlation : Ps ¼ zð1 lnðzÞÞ where z ¼ ð1 Ws Þ (4b)
Equation (4b) tells us that in the case of no correlation, using the example
value for n of 4, and again choosing 0.8 for Ws, then Ps $ 0 ×81, which is a
considerable improvement over the full correlation result which was of 0.67.
Importantly, in cases of an intermediate degree of correlation, the survival
probability would accordingly shift between the two extremes depicted by
equations (4a) and (4b). Put simply, correlation between brain injury and
body injury is bad for survival*the more correlation present, the lower the
chance of survival.
The numerical models and the parameters used in these examples were
not selected to produce a specific results and later we will look at diverse
ranges of values. The main concept being modelled is that the brain can
compensate for minor injuries to the body, and a healthy body can be
controlled by a slightly injured brain, but the combination of relatively
minor injuries to both the body and the brain can be fatal. Therefore
anything that reduces the chance of simultaneous injury of a portion of the
body and the associated control portion of the brain will increase the chance
of survival. In the above example the survival likelihood is higher in the
uncorrelated case, because there is a better chance that, when a portion of
the body is injured, the associated brain portion will not be, and vice versa.
In a way this is the principle behind the familiar adage ‘‘don’t put all your
eggs in one basket’’. Let us now consider how these ideas may be connected
to the question of contralateral vs. ipsilateral control.
EVOLUTION OF CONTRALATERAL CONTROL 333
correlation would not be total, because other injuries would cause harm
mainly to one side of the brain, such as a side impact.
Overall, with four variables, there are six possible statistical correlations
of injury intensity:
(1) left brain with right brain and(2) left body with right body
(3) left body with left brain and (4) right body with right brain (ipsilateral
correlation)
(5) left body with right brain and (6) right body with left brain (contralateral
correlation)
brain will be located on the left side and similarly for right, and in the case of
contralateral control, it will be opposite*the ‘‘control left’’ brain will lie on
the right side. In this case the generalisation of equation (3) is straightfor-
ward and the same for both ipsilateral and contralateral (thanks to this
choice of terminology):
n=2 n=2
n=2 n=2
WL ¼ 1 fCLbrain fLbody and WR ¼ 1 fCRbrain fRbody (5)
Downloaded by [Florida State University] at 10:18 24 December 2014
As for survival, let us assume that, for the animal to survive, both of its
halves must ‘‘survive’’.
The statistical calculation based on these ideas is no longer amenable to a
short-form algebraic expression, but it is simple to model numerically. As
described in Appendix B, a model was developed that allowed for different
scenarios based on a wide range of degrees of correlation of injury severity, a
variety of probability distribution functions for injury magnitude, and
varying degrees of response non-linearity and survival threshold. For each
selected scenario the average survival rate was computed for contralateral
control and for ipsilateral control, based on 100,000 trials*a quantity great
enough to largely eliminate statistical error.
In all scenarios there was a survival advantage for contralateral control.
Over a random sampling of these scenarios, contralateral control provides a
survival advantage of 5%94%. It appears that there are really just two
requirements for contralateral control to be advantageous:
CONCLUSION
Based on the reasoning described here, we believe we have found a highly
plausible mechanism that might have provided the required driving force
behind the evolution of contralateral control. In so doing, we are not
negating the possible existence of other such mechanisms, including those
that have already been proposed by others. One advantage of our new
hypothesis is that it is very general, applying in a wide variety of settings.
This fact is consistent with the very widespread evolutionary adoption of the
somewhat counter-intuitive system of control arrangement. There is plenty
EVOLUTION OF CONTRALATERAL CONTROL 335
of room for further work on this topic. For example, one wonders if the
‘‘injury correlation’’ hypothesis described here could generate new implica-
tions for investigation. For example, the hypothesised greater correlation of
ipsilateral injuries as compared to contralateral injuries may have varying
applicability in different evolutionary settings, possibly depending of the size
of an organism, whether it is an invertebrate of vertebrate, is predator or
prey, etc. Additionally, we note that each side of the cerebellum (the main
balance centre in the CNS) actually controls the ipsilateral side of the body
(in this case via pathways that cross the midline twice). One wonders why this
characteristic evolved, and if this could add further insight to this fascinating
Downloaded by [Florida State University] at 10:18 24 December 2014
puzzle.
Manuscript received 13 May 2012
Revised manuscript received 7 July 2013
First published online 16 August 2013
REFERENCES
Åström, K. J., & Murray, R. M. (2008). Feedback systems: An introduction for scientists and
engineers. Princeton, NJ: Princeton University Press.
Banihani, S. M. (2009), Crossing of neuronal pathways: Is it a response to the occurrence of
separated parts for the body (limbs, eyes, etc.) during evolution? Medical Hypotheses, 74, 741
745.
Bullock, T. H., & Horridge, G. A. (1965). Structure and function in the nervous system of
invertebrates (pp. 5758). San Francisco, CA: W. H. Freeman & Co.
Cajal, R. Y. (1899). Die struktur des chiasma opticum nebsteiner allggemeinen teorie der krauzung der
nervenbahnen [The structure of the chiasma opticum together with a general theory of the
intersection of the nerves]. Leipsig: J. A. Barth.
Doty, R. W. (2003). Unity from duality. Acta Neurobiologiae Experimentalis (Warsaw), 63(3), 163
170.
Drew, J. H., Evans, D. L., Glen, A. G., & Leemis, L. M. (2008). Computational probability:
algorithms and applications in the mathematical sciences (pp. 5565). New York, NY: Springer
Science and Business Media.
Frasnelli, E., Vallortigara, G., & Lesley J. Rogers, L. J. (2012). Left-right asymmetries of behaviour
and nervous systems in invertebrates. Neuroscience and Biobehavioral Reviews, 36, 12731291.
Kirkpatrick, S. (1973). Percolation and conduction. Reviews of Modern Physics, 45(4), 574589.
MacNeilage, P. F., Rogers, L. J., & Vallortigara, G. (2009). Origins of the left and right brain.
Scientific American, 301, 6067.
Rogers, L. J., & Andrew, R. J. (Eds.), 2002. Comparative vertebrate lateralisation. New York, NY:
Cambridge University Press.
Sarnat, H. B., & Netsky, M. (1976). Evolution of the nervous system (pp. 5558). New York, NY:
Oxford University Press.
Serge, S., Raineteau, O., & Jabaudon, D. (2005). Reaching beyond the midline: Why are human
brains cross wired? Lancet Neurology, 4, 8799.
Polyak, S. (1957). The vertebrate visual system. Chicago, IL: University of Chicago Press.
Vallortigara, G., Chiandetti, C., & Sovrano, V. A. 2011. Brain asymmetry (animal). Interdisci-
plinary Reviews: Cognitive Science, 2, 146157. doi:10.1002/wcs.100
336 WHITEHEAD AND BANIHANI
Vallortigara, G., & Rogers, L. J. (2005). Survival with an asymmetrical brain: Advantages and
disadvantages of cerebral lateralisation. Behavioral Brain Science, 28, 575633. doi:10.1017/
S0140525X05000105
APPENDIX A
Derivation of equations 4(a) and 4(b)
Equation 4(a) concerns the case in which there is perfect correlation between
fbrain and fbody in which case they are always equal and both can be
represented as the random variable f having a uniform probability
Downloaded by [Florida State University] at 10:18 24 December 2014
1=n
f ¼ ð1 W Þ (6)
The criterion for survival is that W must exceed Ws, which implies that f
must be less than a value fmax determined from equation (6) to be:
1=n
fmax ¼ ð1 Ws Þ (7)
1
Ps ¼ fmax ¼ ð1 Ws Þ =n (8)
2=n
fmult ¼ ð1 W Þ (11)
The criterion for survival is that W must exceed Ws, which implies that fmult
must be less than a value fmult-max determined from equation (11) as follows:
2=n
fmult-max ¼ ð1 Ws Þ (12)
Downloaded by [Florida State University] at 10:18 24 December 2014
The criterion for survival, together with Equations (10) and (12), yield the
following:
2=n
Ps ¼ zð1 lnðzÞÞ where z ¼ fmult-max ¼ ð1 Ws Þ (13)
APPENDIX B
Numerical statistical modelling of the benefits of contralateral
control
It is possible to simulate different types of injury type correlations through
the use of random variables. This was demonstrated in an Excel spread sheet.
The numerical model employed eight different free parameters, and for each
setting of those eight parameters, 100,000 random trials were run for both
ipsilateral and contralateral control, and the survival fraction in the two
cases were compared.
The first two parameters were survival value Ws and the exponent n as
previously described. In the example described previously these values were,
respectively, 0.8 and 4. In the numerical simulation Ws is randomly selected
with uniform probability density between 0 and 1, and n was randomly
selected with uniform probability density between 2 and 6.
The next three parameters were called C1, C2, C3, and they determined
the relative likelihood of injuries that injured respectively, the whole body,
the left or right side alone, and the head or body alone, relative to the
likelihood of an injury to just one of the quadrants. To determine the
magnitude of injury to the four quadrants, nine independent random
variables, R1 to R9, are determined with uniform probability between 0
and 1 using the Excel function RAND().
338 WHITEHEAD AND BANIHANI
Based on these definitions, formulae for the strength of injury in the four
quadrants are:
TABLE 1
Formulae for randomly determining partially correlated injuries
Left Right
R1 þ C1 R9 þ C2 R7 þ C3 R5 R2 þ C1 R9 þ C2 R8 þ C3 R5
Head
1 þ C1 þ C2 þ C3 1 þ C1 þ C2 þ C3
R3 þ C1 R9 þ C2 R7 þ C3 R6 R4 þ C1 R9 þ C2 R8 þ C3 R6
Body
1 þ C1 þ C2 þ C3 1 þ C1 þ C2 þ C3
Downloaded by [Florida State University] at 10:18 24 December 2014
These formulae yield injury values for the four quadrants, with the
following characteristics: there are correlations between all four quadrants in
proportion to C1, correlations within the left side or right side in proportion
to C2, and correlations within the head or the body in proportion to C3. Of
these, only the parameter C2 is expected to favour contralateral control, and
the model confirms this. In the random selection of scenarios each of C1, C2,
and C3 were selected independently and randomly, with a uniform
probability density between 0 and 1.
Finally, it is desirable to model the possibility that the probability
distribution of injuries might not be uniformly distributed between 0 and 1.
To do this a transfer function was used to convert the uniformly distributed
values described above to various non-uniform distributions. This transfer
function is defined by a formula including two adjustable numerical
parameters, a and b. If both a and b are set to 0, the transfer function
simply returns the input value. Positive values of b shift the output
distribution towards the extremes of 0 and 1, and negative values shift it
towards the central value of 0.5. Positive values of a tend to shift the centre
of the output distribution positively, while negative values do the opposite.
The formula designed to have this effect was:
x
exp a þ expðbÞln 1x
y¼ x (14)
1 þ exp a þ expðbÞln 1x
The average of these values is about 5% with a standard deviation of about 4%.
Overall, this model demonstrates that, under a wide variety of conditions,
as long as the strength of ipsilateral injury correlation exceeds that of
contralateral injury correlation, then contralateral control will yield a
significantly improved probability of survival over that of ipsilateral control.