by Andrew Smart
Given the ubiquity of noise in the brain and the environment, it is not surprising that evolution has endowed biological systems with the ability to use noise to find the signal. In fact if our brains were without randomness, they would not be able to function.
The great thing about our brains is that they have evolved to find signals and truth without any real effort on our part. In fact our brains do a better job of finding our own truth if we are idle.
In the noise field, stochastic resonance (abbreviated as SR) has become an important area of research over the last thirty years. Here’s the revelation: in nonlinear systems, adding a certain optimal amount of noise actually increases the signal-to-noise ratio. In other words, adding noise to a faint signal might actually make the signal stronger.
An Italian physicist at the NATO International School of Climatology named Roberto Benzi introduced SR in the early 1980s to explain the recurrence of the Earth’s ice-age cycle, which happens every one hundred thousand years. This is also the cycle of the eccentricity of the Earth’s orbit. The idea very simply is that there are two “energy wells” or a double well that represent two states of the climate—frozen or warm—that the earth oscillates between.
When the Earth is in one side of the well it’s on average much warmer; when it’s on the other side of the well it’s much colder on average. Benzi postulated that the combination of random or “stochastic” perturbations in the orbit in addition to the eccentricity was what caused the climate cycle; in other words, it was noise. He called the combination of eccentricity and noise “stochastic resonance” to mean that the noise amplified the effect of the eccentricity. In Earth’s case, the source of the noise was small random wobbles in the eccentric orbit that pushed the state of the climate into one state or the other.
Consider the following diagrams:
Imagine that the black marble in the picture represents the state of the climate at any given time.
The wavy line the marble is resting on represents the Earth’s orbit. When the climate is in one of the wells (+1 or –1) it is either an ice-age or warm. When time (t) = 0 in the upper left illustrations, the probability that the climate will jump to the opposite state is very low.
Imagine now that we animate these illustrations and the wavy lines move up and down, and also start to jiggle around randomly. What makes the marble jump from one dip to the other?
The resonance happens when the noise and the orbit combine in just the right way to produce a large change and the marble jumps over the threshold, which could not happen without the noise.
One of the most famous demonstrations in biology of SR came in the 1990s when a group led by Frank Moss at the University of Missouri at St. Louis showed that paddlefish use electrical noise in muddy river water to locate their prey.
Paddlefish feed on plankton in North American rivers. The turbulence and mud make for conditions of near zero visibility. And plankton are tiny. The “paddle” on the paddlefish is actually an electrosensory antenna that responds to the low frequency electrical fields that the plankton emit.
A giant group of plankton causes background noise in the water. Moss’s group found that when they injected an optimal amount of electrical noise into water the paddlefish were able to find plankton that were farther away. This noise enhancement was also demonstrated in the mechano-receptors of crayfish, the antenna of crickets, and in the brains of rats.
Human and animal neurons are nonlinear threshold devices, and as such they actually experience benefits from noise. In fact, it is likely that without noise they would not function at all. When something excites a brain enough, it temporarily changes its dynamics completely. In the case of a neuron it goes from resting to firing off an action potential.
Our neurons communicate with each other via an unbelievably complicated choreography that involves the electrical and chemical coordination of firing patterns among these neurons. Signals travel back and forth, partially synchronizing or desynchronizing their activity as necessary. Each neuron has a dynamic threshold for firing action potentials. In other words, the thresholds change over time. Neurons respond randomly and differently to stimuli, and this response is then randomly integrated to the network to which the neuron belongs.
With around a hundred billion neurons packed into your skull, each firing hundreds of times per second, the inside of the brain is filled with noise. But is this noise bad? It could be that the spontaneous, intrinsic activity of the default mode network provides the necessary background noise for the brain to be able to process information. Abnormal functioning of the default mode network might give you too much or too little brain noise.
Noise can in fact help neurons detect weak signals from the environment or from other neurons.
The figure above shows a typical sinusoidal wave represented by the blue line—aka “the signal.” This could be anything from a sound, an image, a train of action potentials from other neurons, or perhaps even a great poem in your unconsciousness. The dotted line represents the neuron’s threshold for firing.
Note that the blue line never crosses the threshold. Therefore, the solid black line above the dotted line that represents the output of the neuron does nothing. This is a weak signal without noise. It is undetectable.
Now look what happens when we add the right level of noise to the blue signal, represented by the jagged and squiggly red line. Parts of the noise cross the neuron’s threshold (dotted black line) and therefore the neuron fires action potentials, represented by the sold black vertical bars above the black output line.
Notice that where the noise crosses the threshold and causes the neurons to fire, the firing rate corresponds with the frequency of the underlying signal. Therefore the output characterizes the weak signal.
Information is actually transmitted by the noise.
This mechanism also works on the sensory level, so that noise amplifies sub-threshold sounds. Noise can also enhance weak images. A well-known image in the literature on the visual perception of stochastic resonance is Big Ben in London (reproduced from Simonotto, 1998).
On the left, Big Ben is digitized on a 1–256 gray scale at a 256 by 256 pixel resolution.
Each pixel in the picture fires when it crosses a threshold, using the same kind of algorithm as the brain’s neurons. Turning up the noise a little bit by increasing the maximum and minimum random values produces the middle image. This is the resonant noise intensity.
This noise level plus the signal—the weak image of Big Ben—creates the clear image in the middle. The right amount of noise improves the signal to noise ratio. Turning the noise up too high creates the degraded image on the right. When you plot this on a graph, you get what is called an inverted U shaped curve.
* * *
“Were you not always distracted by yearning, as though some lover were about to appear?”
—Rilke, from the first Duino Elegy
During graduate school in Sweden, I studied how noise might help children with Attention Deficit Hyperactivity Disorder, working together with psychologist Sverker Sikström. He developed a model of how stochastic resonance interacts with the dopamine system in the brain, based on the counter-intuitive discovery by psychologist Göran Söderlund that ambient environmental noise actually helps children with ADHD remember a list of instructions. We theorized that noise could be a replacement for amphetamines.
People with ADHD often have very short working memory spans. “Working memory” refers to the ability to temporarily hold information in the brain once it disappears from the environment. Someone rattles off their phone number: how long can you remember the seven digits? Which digits do you remember, and for how long?
We rarely need to engage our working memories very often thanks to mobile technology. However, working memory turns out to be a central cognitive function. If you have poor working memory you’re probably bad at a lot of other things: like time management.
Scientists think the working memory deficit i
n ADHD is related to dopamine function in the prefrontal cortex. Dopamine is one of a family of neurotransmitters that are synthesized by your brain. Without these neurotransmitters, you would not be able to think or to feel anything. This family includes serotonin, norepinephrine, and acetylcholine.
Dopamine underlies many important brain functions like learning, memory, pleasure, and motivation. ADHD children must be very highly motivated to do whatever requires their attention. The idea is that due to genetic mutations that lead to low levels of what is called tonic dopamine (the constant level of dopamine in your brain between your synapses), people with ADHD have an exaggerated “bursty” or phasic dopamine response to internal or external stimulation.
Because the brain is always trying to maintain homeostasis, some imbalance will often be addressed by a compensatory mechanism. In the case of low tonic dopamine, the ADHD brain compensates by releasing a very large phasic dopamine response to any signal.
This is a burst of dopamine, much like the one you get from doing something rewarding like smoking a cigarette, having whiskey, having sex, taking cocaine, drinking wine, eating very expensive chocolate, or of course doing nothing at all. This dopamine rush overwhelms the ADHD brain and it cannot help but focus on it.
For people struggling with ADHD, nearly anything in their environment can cause a huge burst of dopamine. Even more disconcerting, their own internal thoughts and impulses can sometimes cause these large releases of dopamine. In “normal” brains, the extra dopamine released into the spaces between your synapses by something rewarding is then sucked back up so that the balance between tonic and phasic dopamine can be maintained. But the tonic or constant level of dopamine remains high. This allows you to remain focused and motivated.
In an ADHD brain, too much dopamine is being sucked up so that there is very little dopamine in the inter-synaptic spaces and too much dopamine is being released in response to events. Therefore, children with ADHD are hypersensitive to environmental stimuli. This might explain many of the behaviors displayed by children with ADHD: distractibility, impulse control, trouble staying focused, and disorganization. They are constantly thrown between two extremes on a spectrum: cycling back and forth from extreme arousal to complete disinterest.
Amphetamines and cocaine block the reuptake of dopamine and cause more dopamine to be released. Small doses of amphetamine-derivative drugs calm people with ADHD and allow them to focus. Through blocking the excessive reuptake of dopamine in the ADHD brain, these drugs increase the tonic level of dopamine in the brain, while simultaneously reducing the intensity of phasic bursts of dopamine.
Cocaine is pleasurable because it not only blocks the reuptake of dopamine, but it also causes an even bigger flood of dopamine to be released. Over time, the brain stops synthesizing and releasing dopamine on its own because it adapts to having the artificial source.
Without dopamine, life is extremely uninteresting and unrewarding. We do not know at this point what the long-term effects of ADHD medication are, especially on healthy young brains. It is entirely possible that some form of adaption could occur, and less natural dopamine would be produced, which could lead to problems such as depression later in life.
We wondered whether ambient background noise would have a similar effect as amphetamine does for children with ADHD. The idea is that more noise in the environment would allow tonic dopamine in the ADHD brain to facilitate better memory performance. In other words, children with ADHD would require more environmental noise than children without ADHD in order to be able to concentrate.
When we had children with ADHD perform a visual memory task where they had to remember the locations of a sequence of squares on a grid after seeing them for only one second, they could typically only remember the locations of three or four squares. However, while listening to background noise they could remember five, six, or even seven locations, which is the typical visual-spatial working memory span for school-age children.
Using EEG, we saw a dramatic increase in the strength of brain’s response while the ADHD children were listening to noise. The increased neural response could mean that ADHD brains need background noise to stimulate them into working just a little bit harder to perform the tasks of everyday life. The noise, like amphetamine, provides a better tonic level of dopamine that enables children to commit more sustained attention to task relevant information. I suspect that there is a strong cultural and economic component to the increasing rates of ADHD. As the demands of our economy increase in unhealthy ways, the proportion of children who in the past would not have been pushed to the ADHD end of the spectrum are finding themselves unable to cope.
While anywhere from two to ten percent of school age children have ADHD, up to forty percent of prison inmates have ADHD. Children with untreated ADHD have an increased risk for developing drug addiction in adulthood. Likely, these people who might have slipped through the many cracks in our educational and mental health care systems are self-medicating with substances and dosages that can easily hijack their brains.
Interestingly, children with ADHD also show decreased network integrity in their default mode network. It appears that one of the nodes of their default mode network, their precuneous, is not as well integrated into the network as it should be. In a resting state, the spontaneous fluctuations in the default mode networks of children with ADHD appear to oscillate faster than in “normal” children. In other words, these children are actually on a different wavelength. Children with ADHD have a hard time “switching off” their default mode networks. They must work to rest.
Just as noise in the Earth’s orbit helps it to change between climate cycles on a thousand-year time scale, perhaps noise helps the ADHD brain toggle between task positive and task negative networks on a one-second time scale via the mechanism of stochastic resonance. If you have an MRI scanner, some EEG equipment, twenty or thirty children with ADHD, some crackerjack programmers, a free Saturday, a colossal amount of patience, some candy for the children, and some whiskey for the adults, you could probably do this experiment yourself. Let me know how it goes.
* * *
“Hence, it appears timely to turn [noise] from a nuisance into a virtue.”
—Thomas Wellens, noise physicist
Even if you don’t have ADHD, amphetamines improve your memory and concentration while you have elevated dopamine levels. Students have already discovered this, and abuse amphetamine derivative ADHD medication in order to power through ultramarathon study sessions.
We know that people with ADHD tend to be exceptionally creative. This is likely because what is their weakness in a classroom, a boardroom, a cubicle, or a tedious job is really their strength in a music studio, an art studio, a science lab, or an interesting conversation.
To attain the lofty heights of our society, one most possess nearly-psychotic focus. This focus comes at the cognitive expense of being able to see novel relationships among unrelated concepts. Thoughts that are ostensibly irrelevant to what you are doing when you are focused are weak signals from your unconscious that are trying to say, “what you’re doing right now is lame!”
What is bad for time management is good for art. But once you have a creative idea, you need to able to suspend your idea generator in order to focus and get your idea into a physical form. It turns out noise might help you stay in the optimal cognitive range for being creative and concentrating—whether you have ADHD or not.
Recent research by Ravi Mehta, Rui (Juliet) Zhu, and Amar Cheema in the Journal of Consumer Research entitled “Is Noise Always Bad? Exploring the Effects of Ambient Noise on Creative Cognition” found that moderate background noise improved subjects’ performance on the Remote Associates Test (or RAT), which is a well-established test that psychologists use to measure creative thinking.
The RAT is a relatively simple task, which is similar to the “Ten Thousand Dollar Pyramid” game show where contestants tried to get their partners to guess a word without being allowed to say
the word. In the RAT, you are given three or four stimulus words that are in some way related to the “secret” target word. For example, you might be given “shelf,” “read,” or “end” if the correct word is “book.”
Their results show that with moderate background white noise at about seventy decibels, participants were significantly faster at responding to RAT words and gave more correct answers than in low noise or high noise. In other words, moderate noise improves creativity, and a high noise level degrades creativity (as measured by the RAT).
I believe these findings can be perfectly explained by stochastic resonance. I have described how brain regions communicate by synchronizing their oscillations. In this way temporary brain networks are formed to carry out certain functions—like perceiving a scene, listening to a song, or making a PowerPoint presentation.
Through this synchronization, information can spread throughout the network. Adding the right amount of random fluctuation to the system facilitates neural synchronization. Too little noise, and there is not enough synchronization to form a functional network, whereas too much noise can destroy synchronization. Exactly like in the image of Big Ben.
Noise makes the output of downstream neurons synchronize to the frequency of the upstream neurons. At the network level, involving millions of neurons, this noise-induced synchrony mechanism establishes a roughly constant difference between the phases of these weakly-connected oscillators (i.e., neurons). This allows you to have coherent thoughts. Too much synchronization, and you are having a seizure. Not enough, and you aren’t thinking at all.
Lawrence Ward is a neuroscientist at the University of British Columbia who is a pioneer in studying stochastic resonance in the human brain. In 2010, he and his colleagues published a groundbreaking study called “Stochastic Resonance Modulates Neural Synchronization within and between Cortical Sources.” What this means is that noise influences how groups of neurons synchronize their activity within one brain region, and noise also influences how separate brain regions synchronize with each other.