This Amazing Feature of the Brain Lets Us Process Information Even More Efficiently: Dendritic Computation

By on March 15, 2021

Just as “space is really big,” the brain is really complex. You just won’t believe how vastly, hugely, mind-bogglingly complex the brain is. Just take a look at dendritic computation, for example! (FYI, that was a Douglas Adams reference, in case you have never read The Hitchhiker’s Guide to the Galaxy!)

We know that the brain is a black box of mysteries, really. Yet we also have a tendency to oversimplify matters with pleasant metaphors and descriptions. The brain is like a computer. The brain is like a web. Neurons look a bit like tadpoles with a few tentacles coming out the top.

dendritic spines for dendritic computation
Amazing illustrations of neurons and dendrites as depicted by Ramon y Cajal – who discovered dendritic spines!

And if only we could find the inputs and the outputs, we could work the brain like a machine.

In truth, the brain is ever changing. It is in permaflux and infuriatingly difficult to capture. Every brain is different. And even the seemingly most simple action is actually far more confusing that we at first tend to presume.

Introducing: dendritic computation.

Visualizing the Brain

So, just what is dendritic computation?

When picturing the brain, it is common to think of it as a series of nodes, connected by tendrils. These nodes are the brain cells (neurons) and the tendrils are the dendrites that reach throughout the brain. This is what gives the brain its web-like structure: each brain cell (of which there are roughly eighty-six billion) has approximately 7,000 connections (called synapses) to other neurons. The brain of a three-year-old child is estimated to have about a quadrillion synapses, though this number decreases as they get older.

What is dendritic computation?
This classic image of the neuron is an oversimplification

So, yes, the brain does look a little bit like a mind-map. But it is a mind-map of such immense scale as to render the comparison meaningless.

Moreover, the brain is constantly forming new connections and even birthing entirely new neurons. Particularly in areas such as the dentate gyrus of the hippocampus alone (source).

The brain is a mind-map of such immense scale as to render the comparison meaningless.

Then there’s the role of glial cells to consider. These are non-neuronal brain cells that help to maintain and protect neurons. How about long-term potentiation, that strengthens the connections between specific neurons?

Connections run all over the brain, such that some neurons can stretch from one end to the other. This is an important feature of an organized brain and corresponds to the mathematic principle of the “small-world network.”

But that’s not all. Dendritic computation shows us that even this understanding of the brain is sorely limited.

Why We Need Dendritic Computation

Neurons within the brain lie dormant until they receive signals from other cells in the network that stimulates them to fire. This “firing” comes in the form of an electrical impulse called an “action potential” and it tends to correlate with something happening in the brain.

Stimulate a neuron in the motor cortex and you might twitch a muscle. Alternatively, choose a neuron in the visual cortex and you might see a spot of light. Stimulate one in the hippocampus and you might recall a certain memory.

These action potentials travel through the networks of the brain like a Mexican wave. This is why the arrangement and connectivity throughout the “connectome” is so important. When we see one thing, we are reminded of something else relevant. This organization occurs naturally due to one of the simple rules of neuroplasticity: neurons that fire together, wire together. If you keep singing the alphabet in order, then hearing the letter “A” will increase the chances of you thinking the letter “B.”

Elegant, and simple.

Except it’s not!

Synapses… It’s Complicated

Remember: each neuron has 7,000 connections. If action potentials were left to travel freely through that network then they would spread too rapidly and we’d likely experience a sensory overload in response to every single input.

Dendrites

This can lead to a fit or seizure, in fact.

One way this is regulated is through the use of neurotransmitters – brain chemicals that are released during an action potential and that increase or decrease activity in the post-synaptic cell (the recipient). Some neurotransmitters are “inhibitory” while others are “excitatory.”

Each neuron also has a resting potential (around -60mV, in many cases) and must be excited a certain amount before a corresponding action potential will be triggered. Often this threshold potentiation is between -50mV and -55mV but can vary according to many different factors. It might take several inputs from a number of synapses to depolarize the neuron to the required amount such that an action potential occurs.

See also: What is Dynamic Functional Connectivity? Implications for Performance and More

These safety features prevent wanton activity in the brain, but likewise must be finely tuned to enable a rich sensory experience and easy access to relevant information and ideas. Dendritic computation steps in to take this even further.

What is Dendritic Computation?

Essentially, dendritic computation refers to the ability of the dendrites themselves to make basic computations. The dendrites, which are the tendrils that receive signals from neighboring neurons, can actually perform addition, subtraction, multiplication, and division. They can even handle basic logical operations such as AND, OR, and XOR. These are similar to the building blocks used in electronic circuits and computer programming (study).

Logic gate
This is how a logic gate works in a circuit

Far from being simple train tracks that ferry signals to the all-important somas (cell bodies) – a theory known as “point neuron hypothesis” – dendrites, in fact, act more like the complex switches found in microchips.

Dendritic computation is handled by a number of different processes, including groups of membranous “spines” that protrude from the dendrites like smaller branches protruding from a larger branch (reference). These are clustered into “compartments” and that is where the magic really happens.

The role of these dendritic spines is to connect to a single corresponding axon. These spines can act like filters thanks to their narrow necks and bulbous heads that offer additional resistance. They also “space apart” signals to allow for a temporal element.

The Role of Dendritic Computation

This allows for features such as coincident detection: i.e. did two signals come in at the same time.

How might this be useful for our day-to-day lives? It could, for example, be used to detect whether input arrived from two ears simultaneously. The action potential would only occur in cases where that were true (an AND) operation.

dendritic spines

This way, dendrites can consider signals in the context of other signals and they can amplify or reduce the strength of an overall input (reference). We see that voltage signals alter as they travel along dendrites.

The “OR” operation means that a neuron fires if it receives one of two specific inputs but not the other (in the simplest of terms). The 1969 book Perceptrons claimed this was impossible to achieve with a single neuron and some feel this misconception may actually have significantly stunted the progress of AI!

However, we now know that single dendrites are capable of handling this type of computation.

Thousands of these computations operating simultaneously could help us to identify precisely what we’re looking at. For example: enough co-occurring visual cues (size of ears, length of nose, distance between eyes, skin color) could result in the firing of a neuron or neurons relating to a particular person.

See also: Building a Superintelligence: AI vs Exo Cortex

Here is an awesome, in-depth summary of how this is believed to work. It has even been posited that a single neuron could potentially handle enough computation to recognize individual objects!

Something else to consider, is that dendritic computation also mediates synaptic plasticity. These computations mean that certain neurons are more likely than others to connect over time (reference).

(Note that some neurons do work in a more point-like fashion, still.)

What to do With This Information?

So, that’s how dendritic computation changes our understanding of the brain. Now, what do we do with this information?

One thing it clearly demonstrates is the importance of looking after our dendrites and ensuring their health. The reduction of a particular class of dendritic spines, called “thin spines,” has even been linked with cognitive decline in aging.

So, how do we protect our dendrites and support dendritic computation? How do we support dendritic computation?

Glial cells such as astrocytes play an important role in this. We also know that avoiding stress, exercising more, and ensuring sufficient, novel stimulation can all help us to form complex, healthy dendritic structures. Nutrition is also important: a number of specific nutrients may help to support optimal dendritic morphology directly and indirectly. The “satiety hormone” leptin, for example, is linked with rapid, dynamic changes in hippocampal dendritic morphology (study). Increasing fibre intake and supplementing with omega 3 fatty acid (which may also support dendrites through other mechanisms). Conversely, fructose can inhibit leptin receptors and at least one study found a link between high-fat-and-fructose diets and synaptic reduction in rat hippocampi (study).

The concept also presents future areas for research and exploration, it alters our picture of the brain, and it reminds us to be careful of our assumptions. Perhaps it might even force us to re-evaluate strategies such as transcranial direct current stimulation.

But for now, this is primarily an interesting illustration of just how vastly, hugely, mind-bogglingly complex the human brain really is.

About Adam Sinicki

Adam Sinicki, AKA The Bioneer, is a writer, personal trainer, author, entrepreneur, and web developer. I've been writing about health, psychology, and fitness for the past 10+ years and have a fascination with the limits of human performance. When I'm not running my online businesses or training, I love sandwiches, computer games, comics, and hanging out with my family.

2 Comments

  1. nicodemusshai15 says:

    I really enjoy your websites content. thank you keep it up

Leave a Reply

Your email address will not be published. Required fields are marked *

error: Content is protected !!