Asset Publisher Asset Publisher

Return to Full Page

SPECS breakthrough in neuroprosthetics it's a step forward in understanding the brain circuit critical in learning and memory

SPECS breakthrough in neuroprosthetics it's a step forward in understanding the brain circuit critical in learning and memory

After brain damage one could either rely on neuronal regeneration, brain plasticity or on the replacement of the lost tissue by a neuroprosthetic device. The latter have been the theme in many science fiction accounts, but how close are we in building such devices? Can we actually read brain signals and send back information through these devices to restore a lost brain function?


A group of researchers of the Pompeu Fabra University (UPF) of Barcelona, at the laboratory of Synthetic Perceptive, Emotive and Cognitive Systems (SPECS) directed by ICREA  professor Paul F.M.J. Verschure, has taken a critical step forward in replacing a lost function of the brain by presenting a neuroprosthetic system for the cerebellum.

The cerebellum comprises about 70% of all the neurons that make up the brain and is involved in motor and cognitive functions and the processing of time. In particular, Herreros et al. focused on replacing the cerebellar circuit responsible for the acquisition and extinction of motor memories.

specsNEUROPR.jpThe research group has reported the results of their investigation in a paper published in Frontiers in Bioengeeniring and Biotechnology. Bionics Biomimetics,  21 May 2014.

In collaboration with the European Framework 7 project RENACHIP (EU FP7 ICT 216809), SPECS members Ivan Herreros Alonso, Andrea Giovannucci (now at Princeton University) and Paul F.M.J. Verschure, have developed the first neuroprosthetic system that has been demonstrated to replace the learning function of circuits found in the cerebellum in real-time.

To achieve this the SPECS group had to understand the function of the circuit to be replaced, identify its input and output structures, decode and encode the signals generated and read-out by these structures and package all this knowledge into a form that would allow direct real-time interfacing with the living brain.

By tackling all these steps the SPECS team has realized a breakthrough in neuroprosthetics. Their cerebellar neuroprosthesis is based on a computational model of the cerebellar microcircuit involved in the acquisition of conditioned reflexes that SPECS has developed over the last decade.

To turn it into a neuroprosthetic device, this model is abstracted and implemented in a highly optimized fashion, combined with unique compact real-time signal processing methods and trained with data obtained from real-time physiological recordings.

These recordings were made simultaneously from the input systems to the cerebellum (the Pons and Inferior Olive) while the output of the model, a well-timed response acquired by the neuroprosthetic system, is sent back into the brain of the animal through a stimulation electrode. Specifically, the publication shows that an animal can undergo acquisition, retention, and extinction of the eye-blink reflex even though the biological circuit responsible for this task has been inactivated via anesthesia.

With this neuroprosthetic device, or synthetic cerebellum, Herreros et al. have proven for the first time that it is possible to reinstate the function of a disabled brain system by appropriately capturing the underlying computational principles and successfully interfacing the prosthesis with the cerebellar input and output structures. Hence, it also provides direct evidence in support of the underlying computational model developed by SPECS and their overall theory on the operation and organization of mind and brain. [for review see "Distributed Adaptive Control: A theory of the Mind, Brain, Body Nexus". Paul Verschure. Biologically Inspired Cognitive Architectures (2012) 1, 55- 72]

In their paper Herreros et al. also discuss the challenges that such a closed-loop system faces and how their solution provides guidelines for the development of future chronic neuroprosthetic implants. For instance, how a system chronically connected to a living brain has to adapt to the fact that information will not be statically encoded, but instead can be coded by activity whose levels might vary across time and operating conditions.

These results provide further support for the science-based or deductive medicine approach that Verschure and his group SPECS is advancing where medical interventions are based on scientific theory as opposed to correlation or habit. In particular they show how fundamental neuroscience research can drive and inform the development of neuroprostheses to recover brain functions lost due to trauma or aging, first in animals, and in the long term, also in humans.


Ivan Herreros, Andrea Giovannucci, Aryeh H. Taub, Roni Hogri, Ari Magal, Sim Bamford, Robert Prueckl and Paul F. M. J. Verschure (2014), "A cerebellar neuroprosthetic system: computational architecture and in vivo test", Frontiers in Bioengeeniring and Biotechnology. Bionics Biomimetics,  21 May 2014 doi: 10.3389/fbioe.2014.00014