Here on "Earth" organisms tend to evolve only enough mental processing power to handle sensory inputs in modest amounts. Taking "human" brains for a quick example, even though we have millions of sensory receptors all over our body that can feed touch, smell, visual and audio information to our brain, most of the time our brain "tunes out" this information. Clearly, if one was too absorbed in all the rustling of the leaves and scents in the air, he or she would be more likely to remove him/herself from the gene pool by failing to prioritize the important thump thump thump of a charging tiger.

So flipping the script, let's concede right now that this idea seems counter-intuitive on Earth, because it's cost-prohibitive in terms of energy. If a brain is processing things that are not necessary to survive, the plausibility of the species is questionable. But perhaps this is only the case on Earth -- which brings me to my question.


What kind of environment/evolutionary narrative would need to be assumed to allow for high-throughput sensory brains to become a favored trait? That is to say, we are maximizing the amount of sensory inputs that make it to the conscious level.

Further Clarifications:

  • High-throughput sensory brains: All information is preserved and the brain does not "zone out" any "noise." It feels everything from every sensory receptor and passes it to the conscious mind frequently.
  • Number of sensory receptors: assumed to be very high. (hence the post title: high-throughput)
  • Preference: world with a viable food chain in which high-throughput sensory brains have reached high ranks.
  • World: Optional. If you have an Earth-like evolutionary narrative for high-throughput sensory brains, feel free. Otherwise, explain the assumptions of your world.
  • Everything else: only limit is the laws of physics. I will allow for hypothetical biology.

Side: I will include a quote that a rather like as an optional supplement to the post. Originally, I accredited it to Darwin, but I was mistaken. While not widely accepted, it's still an interesting quote.

It's not the strongest of the species that survives, nor the most intelligent that survives. It's the species that responds fastest to change.

  • $\begingroup$ Comments are not for extended discussion; this conversation has been moved to chat. $\endgroup$
    – L.Dutch
    Jul 25, 2019 at 13:27

9 Answers 9


This answer may be voted as non-responsive, but I think your question may be missing your point.

You assert that "filtering out" the unimportant stimulus is a sign of slow or incomplete processing. Instead, I consider it to be a very high-speed approach to eliminating unimportant inputs and allowing the situational intelligence to focus on the unexpected, novel, and important inputs.

Instead of considering the human brain to be slow because it only attends to a small percentage of the signals delivered by the sensors, consider it instead to be very wide, very fast, and well trained at identifying items of importance.

Much, perhaps most, of the work the brain does is not under constant conscious control.

The same brain architecture would work in any situation to identify the important.

If you wanted to make brains faster, you would have a few variables to work with. You could increase the signal conduction rate, reduce the synaptic gaps, and develop faster diffusing neurotransmitters. Brains are already a balanced compromise between size, energy consumption, cooling, waste management, infection control, plasticity, and stability, but any balance can be altered in response to environmental factors.

Improve the sensors.

If you wanted to improve the scope of signals the brain has available, you could make better sensors.

Eyes could have wider fields of view, or be a hybrid of mammalian and insect eyes. Eyes could sense deeper into the infrared or higher into the ultraviolet. Eyes could add sensitivity to light polarization, as some bee eyes are sensitive.

Ears and skin could be more sensitive to vibrations at lower and higher frequencies.

Pattern and texture matching in the visual cortex would evolve to be more responsive to the particular patterns and textures that matter in your world.

The skin could include various chemical sensors for basics like pH and salinity, and specific pheromones important to your creatures.

What type of world would encourage this?

To encourage better vibration sensing, a world with large threats that shake the ground.

To encourage a polarization sense, a world with a fairly thick atmosphere but little variation on the surface, perhaps an ocean or desert world.

To encourage infrared sensing, a cold world with warm-blooded threats.

To encourage an ultra-violet sense, a blue star and little ozone in the planetary atmosphere.

To encourage chemical sensing, a dim world, perhaps underwater near a hydrothermal vent.

To encourage wider field-of-view, a world where threats come from all directions. This could be creatures who live on the surface and suffer threats from the air, or perhaps an avian or underwater species.

  • 3
    $\begingroup$ Don't sell yourself short. This is an excellent answer, and very much on point. $\endgroup$ Jul 23, 2019 at 19:38
  • 1
    $\begingroup$ Having worked on many high throughput data systems, I have to agree. Take for example the Compact Muon Solenoid generates 40TB/s. It uses a series of L1-L5 triggers to filter out events, giving an overall output rate of about 1GB/s, even that takes one of the largest compute grids in the world to process. $\endgroup$
    – Aron
    Jul 24, 2019 at 6:59
  • 1
    $\begingroup$ @VLAZ Not just the brain. For example the eyes themselves, through interconnected reenforcement/dampening generates pretty good edge detection, which you can see in action with optical illusions, such as the famous grid/dot illusion. $\endgroup$
    – Aron
    Jul 24, 2019 at 7:02
  • 1
    $\begingroup$ The first paragraph raises a very valid point (input is processed, just not consciously). However, the following discussion about speeding up the brain and adding more types of sensory input seems a bit of a non-sequitur. How about addressing (instead of, or in addition to what's already written) the conditions for encouraging the selection of brains where the sensory input is accessible to the conscious mind? $\endgroup$
    – G0BLiN
    Jul 24, 2019 at 14:45
  • 1
    $\begingroup$ "...skin could be more sensitive..." which is where hair comes into play. It can help detect very high and very low vibrations. A hairless arm doesn't detect a change in the wind like a hairy arm will. $\endgroup$ Jul 24, 2019 at 16:41

I think it's more than just the world the creature lives in you also need to account for their habits and the their place in the ecosystem to that end I believe you need to start with a warm-blooded obligate carnivore, such creatures are almost always on the hunt for the next meal as it is. Then you need to put them in the middle of the foodchain rather than at the top. Now they're either on the look out for a meal or, if they have a full belly, the things that hunt them.

Next to maximise the sensory data that they have to process while on the lookout have all local herbivore populations be toxic at certain stages in their life cycles and the life cycles of the vegetation they feed on. Predators now need to take into account the tiny signs around them in the vegetation and on their prey to know what meat they can hunt at any time. Furthermore have the higher order predators that eat them take full advantage of active, chameleon-esque camouflage, so they're harder to spot at a distance and be adept at both stalking and ambush hunting tactics.

So now we have a creature that is constantly scanning its environment for, prey suitability cues, prey, and the stealthy higher order predators that specifically prey on it, no detail can be safely dismissed.

  • $\begingroup$ I'd wager that in such an environment, the brain would be even more attuned to very specific sensory cues, and completely ignore the rest. The creature wouldn't be extra-aware of everything, but blind to anything that isn't directly part of the food chain. $\endgroup$
    – Sheraff
    Aug 13, 2019 at 16:21
  • $\begingroup$ @Sheraff The point of the environment in question is that absolutely everything is directly part of the foodchain. $\endgroup$
    – Ash
    Aug 13, 2019 at 16:24
  • $\begingroup$ then you also need for the "food chain" to have no identifiable pattern. If anything repeats reliably enough that it can be used to identify food or predator, then that's what the brain is going to use as a shortcut. Think of the brain as a deep-learning AI tied to an array of sensors. Given enough samples to train from, it is eventually going to under-use some data and over-extrapolate from some other. $\endgroup$
    – Sheraff
    Aug 13, 2019 at 16:28
  • $\begingroup$ @Sheraff There are individual patterns but think about cicadas and how often the 13 and 17 year varieties come out in the same summer, once every 221 years. Now multiply that kind of occurrence across dozens of species of plant, prey, predator, and super predator that are all interacting. $\endgroup$
    – Ash
    Aug 13, 2019 at 17:07

Scarcity of resources is a fact we have to deal with, no matter if we are someone with an IQ of 200 or an amoeba in a muddy puddle.

From scarcity of resources follow an economy principle: you use something only if the cost is justified, else it is just a waste of scarce resources. So, for example, if you are an organism living in an underground lake, eyes are pleonastic and will be removed by evolutionary pressure, based on the above economic principle.

Coming to your question, if you want the "background noise" to be processed and not ignored, it doesn't have to be a noise, but meaningful and valuable. This can be true for a limited subset of signals in a limited set of circumstances (i.e. a goal keeper during a match might be way more reactive to someone else's movements than he is while standing in a queue at the post office), for the rest it makes hardly sense.


Most of the information filtered out is next to useless. For instance, if you have ever seen "Tracers" in movies where they are showing people taking drugs, that's an actual thing, not a hallucination--your eyes do not have perfect resolution or response speed. When something moves in front of your eyes it takes a while to fade--your brain figured this out when you were a baby and built a filter for it. Your current vision without the tracers is the actual hallucination!

If your conscious is always presented with the 1000 useless smells that are in your home, you might have trouble picking out the new smoke smell informing you that something bad is going on, and if you are always tracking every single follicle your shirt touches you might not feel a wasp that just landed on your arm and the pain from a pin prick might be debilitating for days.

We probably make nearly optimal use of our senses as they are. To have better senses that could actually pick up additional information you might need a nose like a beagle or physically larger eyes.

If you want to try processing all the sensor input manually, there are certainly ways to experiment with that.


Your question comes very close to answering itself. For the evolution of brains to be driven by high levels of sensory information this would have to take place in an environment that would be extremely complex, extremely active with lots of stuff happening, and extremely hazardous. In other words, the processing of sensory information becomes the major imperative selection pressure.

Such an environment might develop on the type of planet referred to as a super-habitable Earth. Planets of this type could have the necessary complex, active and hazardous environments to drive brain evolution through sensory processing.

Environments of super-habitable planets have the potential for the evolution of overabundant ecosystems complete with extreme numbers of predatory and competitive species of animals and plants. That's just the biota at the macro-scale, such super-habitable planets would have ferocious and deadly microbiomes. Evolving the ability and capacity to detect colonies of hazardous micro-organisms could be essential. To say nothing of clouds of spores and seeds capable to infecting animals and plants.

Any organisms developing brains of any size will favor the evolution of massive arrays of sensory organisms and due to these selective pressures this will in turn drive the evolution of their brains. Those brains will have correspondingly high levels of sensory processing.

In conclusion, it's purely a matter of understanding what sort of environments which produce the necessary selective pressures to drive the evolution of brains in relation to their sensory capacity.

  • $\begingroup$ The problem is that responding to threats requires speed, not necessarily depth of thought. While the organisms might have heightened senses, their priority would be low latency rather than high throughput. (In a way, humans are already built like this. We use a lot more of our sensory input than is consciously available to us.) $\endgroup$
    – Cadence
    Jul 23, 2019 at 8:23
  • 1
    $\begingroup$ @Cadence My answer is focused on sensory information processing, not depth of thought, in an extremely complex, active & hazardous environment. Many organisms have sensory mechanisms to alert them to something they need to be altered to. For example, something that moves or is the right color for ripe food. Consciousness doesn't need to be part of the picture. The same way it is in humans and other organisms. So we are in agreement. $\endgroup$
    – a4android
    Jul 24, 2019 at 5:31

I think you're missing a few details of how brains work, so I'll jump in with a neural network:

diagram of neural network, from Wikipedia

You say...

I kind of just made this term up, so let me explain it briefly here. All information is preserved and the brain does not "zone out" any "noise." It feels everything from every sensory receptor in its whole body all the time.

Here's an interesting fact: a neural network processes all the input, all the time.

At a theoretical level, a neuron is something that takes several inputs (between zero and one), and outputs the sum of its inputs, giving more importance (weight) to some inputs than others. At a rough biological level, a neuron is a cell that takes inputs from either sensors or other neurons in the form of electrical charge (via chemicals called neurotransmitters), computes the sum of the inputs via chemical reactions, and "outputs" an electrical potential.

Now, you might argue that "zoning out noise" is setting the output of any neurons at the middle layers to zero, and you never want that. That would be a self-defeating argument, as you would need a minimum value for the neuron output, which would become the new zero. Any neural network is gonna have neurons with minimal output at any given time.

That said, I think you are aiming for a wide neural network (lots of neurons as input and lots of neurons in each layer) as opposed from a deep neural network (lots of layers). Let me quote from https://stats.stackexchange.com/questions/222883/why-are-neural-networks-becoming-deeper-but-not-wider :

The main issue is that these very wide, shallow networks are very good at memorization, but not so good at generalization.

You can, in fact, have a very fast neural network (since less layers means less propagation time), but it's not gonna remember anything, and making that neural net learn anything is gonna be a long process.

You would also have a problem with memory, since it depends on loops in the network.

Then you can say "well, let's just add more neurons to remember everything, let's make the network both wide and deep"! to which the response would be: such an organism would have an evolutionary disadvantage due to energy usage:

Most vertebrate species devote between 2% and 8% of basal metabolism to the brain. [...] in humans it rises to 20–25%.

In other words, such a creature would need to consume more food than a stupider counterpart able to perform the same abilities.

So let me flip your question and try to give a world-building answer:

What kind of environment/evolutionary narrative would need to be assumed to allow for brains with wide neural networks to become a favored trait?

I'm gonna say:

  • No need for long-time memory
  • Rely only on instinctive responses
  • High mortality and fertility rates
  • Reduced metabolic cost of sensory input organs

So, Zerglings, but with weird fly eyes.

  • 1
    $\begingroup$ I really like the actual answer at the end of this post - maybe expand on it a bit? At the moment it's just a list of bullet-points... $\endgroup$
    – G0BLiN
    Jul 24, 2019 at 14:54

I believe it will be a very slow paced environment where resources are very rare.

If your resources are hard to come by, you need to invest into a few select sensory organs to help you look for that next meal. This means you specialize in processing all the details that a sensory organ can provide you. So as a twist to your question, you do end up processing all the noise you can, but your noise is limited because you have limited detection methods.

For example, if a creature only has a sense of touch, because the live in a super dark environment with no sight, no ability to smell and no ability to hear, they can only focus all their attention of finding out what has touched them.


Automated sensory selectivity, i.e. not passing sensory data to the conscious mind, is the result of evolution. Certain sensory inputs have been deemed irrelevant and thus placed in a sort of second tier data, which is passed to the conscious mind only if you focus on it.

Other answers have provided the obvious explanation that this effect is due to an economy of energy and the need for quick responses to environmental anomalies, which could indicate presence of predators, immediate danger, or simply a new and unknown situation.

To prevent this filtering to happen before conscious processing, you may want to avoid any input to be associated with anything usual. I will call this the Schrodinger world, whereby it is impossible to know from environmental cues alone what situation we are in, e.g. whether of danger or safety.

As no specific combination of sensory data can be disregarded over other combinations, two evolutionary parts are likely:

  • energetically cheap: no sensory data is processed.
  • energetically expensive: all data is passed to the conscious part of the brain, which has to add non sensory information to decide.

It is a bit complicated to imagine a scenario in which all sensory data is not sufficient to determine the chance of a situation being free of danger, and yet not being completely useless.

To allow such scenario I'd go with peacock-humans, who display all their sensory appendages in a world where they serve no purpose, only to gain a mating advantage: perceiving more of the world is also being perceived more. Mating among such humans is a conscious battle in the entire sensory spectrum, like a chess game in which the defeated candidate retreats after realizing they have no valid responses to the opponent nuances.

Why wouldn't this mating dance be handled by the subconscious over evolutionary time scales? The ever changing nature of the Schrodinger world is such that repeatable behavior is penalized. A predator may be lurking for mating humans that display the same patterns over and over, thus keeping their numbers dwindling, as opposed to the very successful conscious-peacock humans.


In order to increase the amount of relevant details an organism pulls from its sensory data you would need an organism to live in an environment that requires it to know about more details about its environment to survive.

To maximize the details the organism needs you need it to be a predator but not an apex predator. This requires it to use its senses to overcome the hiding techniques of its prey and it requires the organism to monitor its surroundings for incoming threats.

Next you need the organism to be in tune with its surroundings to move. For example, if the area the organism lives in is flat and stable it requires a minimum amount of relevant details to move about in a predictable way. On the other hand, if the organism moves in a volatile medium, say it walks, but on mountains with a lot of seismic activity, it fly's in turbulent winds, or it swims through chaotic currents the organism will need to know a lot more about its environment to safely and precisely move about.

The organism would also require additional knowledge about its surroundings to survive if it used camouflage for multiple senses as its primary means of hunting and staying safe. If the organism camouflages its self visually, it would need to be able to pick out details better than its prey to attack and better than its predators to stay hidden. If it camouflaged its smell it would need to be able to smell better than its prey and its predators as well, etc.

Lastly the organism would need to be warm blooded with a fast metabolism. This increases the energy requirements for the animal and would force it to actively move about which would increase the number of times per day it has to rely on subtle sensory information to survive.

I imagine something like an arboreal shrew that glides between trees like a flying squirrel and has camouflage skin like an octopus. It would need to sense the wind velocity on its skin, the kinesthesic positioning of its body, the velocity of the branch its standing on, the position of the branch it would need to leap to the smell it is trying to duplicate as part of its camouflage, the colors/textures/patterns of the branch it is camouflaging in, all just to determine the right moment to pounce on prey or to flee to another tree.

In an environment that has many effective predators and quick elusive stealthy prey the organism would have to be aware of hundreds of tiny details to avoid death and would definitely have the high "throughput" senses you were looking for.


You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .