Where is Memory: The Brain-Wide Network

A fundamental question that fascinates me is if the brain is unified in its perception of the world. Or is it just the sum of separate parts, like a computer?

The cerebral cortex is very plastic and can adapt to a wide variety of functions, dynamically building areas for specific functions [1]. This plasticity extends through the entire human lifespan, so the it is fundamentally a dynamic and adaptive organism. So we see the brain has the physical abilities to form a decentralized network.

In another view of the brain, it is set up like a computer; separate parts such as the visual cortex and the motor cortex connected to the sensory organs and some sort of Central Processing Unit (CPU). So we should be able to find the “hard drive” where all the memories are stored, and a tweak fixes it and cures Alzheimer’s. Just as fMRI’s can scan the visual cortex and infer what picture a subject is looking at [2], we can scan the hard drive and lift memories.  This explains why a hit to the head can cause extensive amnesia, the hard drive is knocked out.

However, we have not been able to find such a centralized hard drive. In fact, rats who learn to navigate a maze can still do so, after we remove up to half of their brains [3]. This suggests that memory is stored across a distributed network spanning the entire brain. Extensive trauma to the brain often does not critically impair memory or higher order thinking, because the brain-wide network is still basically intact.

This is damning to the view that the brain is the sum of parts. If memory persists through the removal of 90% of the brain, it seems even more unlikely that there is a part from which consciousness arises. This sort of God module seems like a pipe dream of reductive materialism, when in fact consciousness is just distributed throughout the brain, without a central processing unit to direct it.

How does this network function? We don’t remember our infancy. We haven’t formed mental building blocks; basic sensory qualia such as the color red or the shape of a hand. Memory could boil down to connecting and associating these blocks. We only remember things for which we have building blocks, and without those blocks, the network does not form [4].

Similar to our memory-less infancy is the case of the “snap trip” in psychedelic experience, where the subject finds himself jumped hours forward, without even the vaguest sensation of passed time. The psychedelic experience may be composed of no recognizable building blocks, leading to a situation like infancy. Without basic qualia blocks, the network of memory does not form. In this view, time itself may just be another qualia.

This decentralized network would have to span the entire brain. Visual qualia exist in the visual cortex, emotional qualia in the amygdala, and movement in the motor cortex. These are located in entirely different regions. So memory is a physical network between these different regions. Even if one region is knocked out, the network still persists and memory is preserved.

More to come on the brain as a distributed, decentralized network…

1. “Patterning and Plasticity of the Cerebral Cortex” by Mriganka Sur, John L. R. Rubenstein. http://science.sciencemag.org/content/310/5749/805
2. “Scientists use brain imaging to reveal the movies in our mind” by Yasmin Anwar. http://news.berkeley.edu/2011/09/22/brain-movies/
3. “Karl Lashley”. http://www.newworldencyclopedia.org/entry/Karl_Lashley#Work
4. Late night discussions with Austin Thornbury

2 thoughts on “Where is Memory: The Brain-Wide Network

  1. Suppose we have forms of energy in spacetime.
    My experiencing of time is defined as my experiencing of variance and invariance of these forms.
    Invariance is hard to define: perhaps a form is invariant when an invariant experiment on it produces invariant output

    Given a frame, memory refers to the invariance of form.
    In humans, there is a delocalization of memory where each cell possesses a copy of the same DNA sequences.
    The DNA acts as a deep human memory, of which each cell is conscious.

    The physics of gravity and electromagnetic forces, and the physics of the observer effect show that interactions are bi-directional; to observe something is to act upon it.
    Consciousness of something is interaction with it. Interaction is a symmetry in the variance/invariance of two things that identifies the pair as one thing.
    E.g. a brick demonstrates its consciousness of the earth by falling towards it.

    In a network, a set of objects interacts with one another. Given interaction between a pair of objects, we cannot fully decouple the pair, thus we must regard the whole network as a single object.
    However, some subsets of these objects interact more significantly together/internally than with external objects, thus forming modules.
    Given a pair of modules, perhaps we can try to study the internal interactions, each independently of the other.

    Is there a modularity of particular human-consciousness functions?
    I’m not sure, but I imagine there could be. Perhaps I can develop my tennis skills completely independently of my abstract mathematical reasoning, if the two conscious functionings occur primarily in self-contained patterns, with little overlap, each with its own addition functions. On the other hand, perhaps the two functionings share common basis functions (such as the addition functioning), and I cannot develop one without developing the other.
    Chomsky argues that the human language competency exists in a separate module, rather than as the emergent property of deeper and deeper interconnectivity of the processes in general.

    Is there localization of each human-consciousness module to a particular area of the brain?
    I could imagine that each module is distributed throughout different areas of the brain, such that any area of the brain can contain the processing/memory of multiple modules (either overlapping, or separated internally) E.g. you can arrange it so that things that are contiguous wholes in virtual memory are fragmented in physical memory.
    I would imagine that while there may be multiple copies of a memory, if all copies of that memory are deleted, then in that sense, the memory was localized and removing those localities removed the whole memory.
    I would imagine there would be some sort of central processing, that synthesizes memories, that this kernel would be at a constant location.
    I’ve heard of localized areas of the brain that correspond to particular functioning, like the hippocampus for memory, and the amygdala for emotions, and Broca’s area and Wernicke’s area.
    perhaps some area can be assigned to new conscious functioning; this doesn’t seem rule out the possibility of modularity of conscious functioning

    If I were to imagine the general human consciousness, I see modularity in the denser interconnectivity of the parts of an individual’s mind than between different individuals. I also see this plasticity, where new minds replace old minds, with the same conscious functioning, or an individual mind can learn new things. I think if a region is knocked out, the memory in that region is not necessarily persisted by the memory of the network, unless that memory has duplicates that haven’t been destroyed, or the rest of the network has the plasticity to reproduce it.


    1. Thanks for your thoughts Peter, I couldn’t have said it better myself!

      The brain probably lies between the two extremes of unified network and modularized machine. Let’s think of it as a network of modules. As you mentioned, some of these modules are “kernel nodes”, which play a critical role in the synthesis of some function such as memory. However, these kernel nodes do not define memory, simply because its removal would critically impair it. Memory emerges from the entire network. It depends on building blocks such visual and olfactory information, information in the network as a whole. We cannot point to the kernel and say it is memory. It is merely a highly connected, critical node in the network. Yet the memory kernel in isolation would not give rise to memory.

      Similarly, the language kernels may be located at Broca’s and Wernicke’s areas. However language would fall flat if it could not form associations with sensory experience and memory. Language is similarly an emergent property of the entire network. We cannot isolate the kernels and say it is language.

      Although modularity is a critical feature of the brain, the network behavior is what really gives rise to the salient functionality of consciousness.


Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s