Last updated: February 19, 2013

Neurona@Home is a project that takes hundreds of internet-connected computers from all over the world and uses them to simulate one million neurons. Uses the BOINC distributed computing platform to simulate a 1,000,000 neuron bee brain Two types of neurons: inhibitory and exitatory Each neuron can be one of three states: resting, activated, refactory Model assigns each neuron a state and chooses interconnection values

BOINC distributed computing platform to simulate a one-million-neuron brain of the honey bee. Two types of neurons are simulated: inhibitory and exitatory. Each neuron can be one of three states: resting, activated, refactory. The model assigns each neuron a state and chooses interconnection values. Lead researcher is Dr. Javier Villanueva at the Complutense University of Madrid.

Contents    (hide)

Latest news / current status
 Neurona@Home finishes calculations for now, all WUs have been processed.
 Time will now be taken to analyse the results.

results of phase 4 - 22 may 2012

results of phase three - 10.Mar.2012

Project started simulations - 14 June 2011

Technical details

Towards the end of the simulations nearly 2,000 computers (known as "hosts") were taking part.
aim is simulate the behaviour of a large assembly of cellular automata neurons connected
in a complex network

Human brain contains ~86 billion neurons with an average of 10k synapses per neuron.
Honey bee brains contain ~960k neurons, gives them memory and behavioural repertoire

-	It is known that the synchronization of the firing of large assemblies of neurons
is related to odour discrimination in insects.
-	In Humans synchronous oscillations are also present but they are usually related
to higher-order functions such as attention, memory and conscious awareness.
-	The origin and even the role of these oscillations are still poorly understood.

We consider a model in which a set of excitatory and inhibitory nodes can activate
or deactivate neighbour nodes according to a set of probabilistic rules. The cellular
automata neuron is a very simple artifact with only three states: firing, resting or
refractory but we are interested in the features of the brain that emerge as a
consequence of its complex network structure.
-	To this aim we should consider an artificial brain with 1.000.000 neurons, a number
that exceeds the typical size of an insect brain.

Our project's simulations of the cellular automata brain in a random network should
allow us to understand:
-	The emergence of brain oscillations as collective firing.
-	The effectiveness of artificial neural networks as discriminant devices.
-	The role of the network structure of brains in information processing.

Each node or neuron can be simulated with three states:

(i) neuron at rest, one which is capable of being excited by the neighbours in the
network structure to which it's connected.
(ii) active neuron, one which is emitting electro-chemical impulses through its
(iii) refractory neuron, one which has recently stopped firing and must undergo
a period of time before returning to idle.
Resting neurons can be excited with a certain probability by active neurons in the
previous time step.

Although our model of simulated mini-brain neurons has the same number of neurons
as the one of a bee, we do not intend, of course, to explains the entire repertoire
of behaviours of a real bee. Our goal, more modest, is to explain the collective
activity typical of the neural network.

It is known that brain oscillations, quasi-periodic behaviour in rhythmic brain
electrical activities, are present from insects to humans. In the case of bees, locusts,
or the fruit fly, oscillations in connection with the encoding of odour have been found .

In the case of human beings, the purpose of these oscillations appears to have been
shaped by evolution. Theta waves appear to be associated with memory, attention and
even consciousness.

These oscillations allow insects to discriminate between very similar odours.

This project will explore the parameter space of the model to determine those regions
that could correspond to the synchronous activity of the brain and explain its role in
odour discrimination tasks (in insects) or memory and attention (in humans).

For each Work Unit the model provides a network of 1,000,000 generated neurons with a
specific degree of connectivity k. It uses two types of neurons:

inhibitory neurons 
These neurons, when active, have the effect of inhibiting the excitation of a resting
neuron that could be activated by a neighbouring excitatory neuron.
excitatory neurons 
These neurons, when active, are able to trigger (excite) their resting neighbours.
The model assigns to each type of neuron one of the above three possible states
(resting, activated or refractory) and chooses a range of interconnection values
between neurons, and then it checks if these values provide a synchrony in neuronal
activity and therefore if the suggested model is feasible or not.

============ Mathematical models

So defined, our model shows similarities with the SIRS model
(Susceptible-Infected-Recovered-Susceptible) Mathematical Epidemiology, where resting
neurons play the role of individuals susceptible to disease, active neurons are
infected individuals and refractory correspond with recovered.

It is quite common in Applied Mathematics that the same model developed with very
different goals naturally emerge in another completely different field of research.

Experience in BOINC projects for epidemic (Respiratory Syncytial Virus) allows us
to build up a model to study the propagation of signals in a simulated brain. The
mini-brain study will have a million neurons which puts it in the range of
advanced brains among insects.



The Berkeley Open Infrastructure for Network Computing (BOINC) is an open source middleware system for volunteer and grid computing. It was originally developed to support the SETI@home project before it became useful as a platform for other distributed applications in areas as diverse as mathematics, medicine, molecular biology, climatology, and astrophysics. The intent of BOINC is to make it possible for researchers to tap into the enormous processing power of personal computers around the world. BOINC has been developed by a team based at the Space Sciences Laboratory (SSL) at the University of California, Berkeley led by David Anderson, who also leads SETI@home. As a high performance distributed computing platform, BOINC has about 540,130 active computers (hosts) worldwide processing on average 7.296 petaFLOPS as of December 2012. BOINC is funded by the National Science Foundation (NSF)

Further details on Wikipedia: BOINC -

How to participate

How to take part
 You will need a 64 bit Linux or Windows PC machine with suitable amounts of RAM
 Each work unit takes 3 to 6 Gigabytes 
Download the client:
Send an email to Javier Villaneuva (address) or leave a message in the comments box on this page:
There were 600 machines initially in August 2011 and this grew to nearly 2,000 by the end in February 2012

Future research

Simulation efforts completed in February 2012.
Work is now continuing to analyse the results.
There are no plans to develop the research further.
Researchers will instead use same applied maths, and same distributed computing,
to model spread of menincoccal disease.

Previous similar projects

This was a previous similar project that closed in 2010:

People involved

L. Acedo -- Luis Acedo
J. Villanueva-Oller -- Javier Villanueva-Oller
J. A. Moraño -- José Antonio Moraño
R.-J. Villanueva -- Rafael-Jacinto Villanueva

Javier Villanueva (redirects to using frames)

Lead researcher: Javier Villanueva at The Complutense University of Madrid

Professional email:


FIS PI-10/01433
PAID-06-11 Ref: 2087
  • Homepage:
  • News forum:
  • Message board:
  • Paper: - requires payment