IBM Designs Chip That Replicates Human Brain

The human brain might soon be replicated in a chip. IBM hopes to replicate the product of at least 7 million years of evolution and create a chip that is based on the model of the brain and functions like one. This is a new direction in chip making, going for diverse functionalities and simulation of logic centers than mere parallel processing and speed.

A Big Announcement

IBM announced today that, along with four universities and the Defense Advanced Research Projects Agency (DARPA), it has created the experimental computer chip that mimics human brain processes. The main goal for the collaboration is to create a chip which will be able to perceive the environment, prioritize goals, interact with the surroundings and produce a proper response. It will also be able to increase the efficacy of the response process by basing later responses on the outcome of the previous responses and their consequence a sort of artificial intelligence, but at the level of a human brain. The DARPA project is called Systems of Neuromorphic Adaptive Plastic Scalable Electronics or SyNAPSE.

The Chips

The speed of the chips is really slow too slow in comparison to the modern chips made by IBM itself. The processing speed is a paltry 10 Hertz, but this is one chip not looking for speed. The human brain is a wonderful device for parallel processing on an unthinkable scale. Moreover, it is able to create banks or sites, which can act as logic or memory locations. These junctions between neurons synapses have also been modeled on the chips.

The brain model

The current devices our known laptops or desktops are all von Neumann devices, which means that they can process information at a speed determined by how fast data can be carried by a bus. Higher capacity and speed of the bus means that the data is processed at higher rates. The speed of the computer is thus limited by the capacity of the bus, a phenomenon called von Neumann Bottleneck’.

According to Dharmendra Modha, the head of the DARPA project and associated with IBM, Almaden, the chips have deviated from von Neumann behaviour:

 We are now doing a new architecture. It departs from von Neumann in variety of ways.

These chips are built for parallel processing on an unprecedented scale. The memory center has been achieved by a small conventional memory chip, an incorrect description of how memory works in the human brain. This is mainly due to the fact that we know very little as to know synaptic nodes can act as centers of correlated memory.

Dharmendra Modha. He was the principal researcher behind the DARPA project. He's associated with IBM Almaden.


The crucial test for these chips will come when they will be called upon to do tasks that require genuine inspiration or any other quality that we associate uniquely with the human brain. Can these chips make mistakes? Or forget things? Or remember something related but not the exact fact? Can it discover? Can it feel inspired? These will be future questions as this is only the beginning. Jeopardy champion, Watson will have to pale in comparison!

Maybe, in some time, you’ll find the current author of this article replaced by a chip, a few cm across on either side, writing about its own history.

Published by

Debjyoti Bardhan

Is a science geek, currently pursuing some sort of a degree (called a PhD) in Physics at TIFR, Mumbai. An enthusiastic but useless amateur photographer, his most favourite activity is simply lazing around. He is interested in all things interesting and scientific.