Your ears make sound, and nine other measured facts about your nervous system

Your ears make sound

Healthy ears give off faint sounds of their own, and a small microphone in the ear canal can record them.

In 1978 David Kemp played very quiet clicks into human ears and recorded what came back. For the first 5 milliseconds or so the ear simply echoed the click. After that, in every normal ear he tested, a second response appeared that faded far more slowly: its decay time changed from about 1 millisecond to more than 12. It was absent in ears with cochlear deafness.

The sound comes from the cochlea. Kemp’s later review describes these otoacoustic emissions as sounds caused by the motion of the cochlea’s sensory hair cells as they respond to sound, and notes that recording them is widely used in newborn hearing screening programmes.

Some ears also produce a faint tone with no sound going in. In one study of adults, Penner and Zhang found these spontaneous emissions in 83% of the women and 62% of the men tested, after analysing each recording 15 different ways and combining the results. How the recording is analysed changes the answer: any single analysis found them in 64% to 81% of the women and 39% to 55% of the men. The authors say this helps explain why earlier surveys disagreed.

Kemp, “Stimulated acoustic emissions from within the human auditory system”, J. Acoust. Soc. Am. 64, 1386 (1978); Kemp, “Otoacoustic emissions, their origin in cochlear function, and use”, Br. Med. Bull. 63, 223 (2002); Penner and Zhang, “Prevalence of spontaneous otoacoustic emissions in adults revisited”, Hear. Res. 103, 28 (1997).

How it was checked. Abstracts of all three papers. The full texts are behind paywalls and were not opened, so the sample sizes and methods in Penner and Zhang are not reported here.

A stubbed toe can hurt twice

One sharp injury sends signals along two kinds of nerve fibre. The fast ones arrive first and the slow ones much later.

Pain fibres come in two main kinds. Aδ fibres, wrapped in myelin insulation, conduct at about 15 metres per second. C fibres, thin and unwrapped, conduct at about 1 metre per second. A single intense stimulus to the skin therefore produces a pricking “first pain” followed by a burning “second pain”. Thomas Lewis and Edward Pochin published a study of this double response in 1937.

Over a short distance the two signals arrive almost together. Over a long one the gap grows. Drag the slider to change the length of nerve between the injury and the spinal cord.

Jin et al., “Somatotopic representation of second pain in the primary somatosensory cortex of humans and rodents”, J. Neurosci. 38, 5538 (2018), introduction, which gives the two speeds and cites Lewis and Pochin, “The double pain response of the human skin to a single stimulus”, Clin. Sci. 3, 37 (1937).

How it was checked. Full text of the 2018 paper’s introduction. The 1937 paper was not opened; it is cited only as the paper that the 2018 authors cite for the double sensation.

Your retina is wired backwards, and glial cells help light through

Light has to pass through several layers of nerve cells before it reaches the cells that detect it. Glial cells that span those layers appear to carry it across.

The vertebrate retina is inverted with respect to its optical function. The light-detecting cells sit at the back, behind the nerve cells that process their signals. Any image sent through layers of randomly oriented cells should be scattered and blurred.

In 2007 a team examined the retina’s Müller cells, glial cells that span its whole thickness. They have a funnel shape and a higher refractive index than the tissue around them, and they point along the direction of the light. Confocal microscopy of retinal tissue showed a low-scattering passage from the surface to the light-detecting cells, and a laser trap showed individual Müller cells guiding light like optical fibres. The authors compare the parallel array to the fibre-optic plates used for low-distortion image transfer, and write that the cells “seem to mediate” the transfer.

Most of the experiments used guinea-pig retina. The authors report the same reflection pattern in rabbit retina (data not shown) and in human retina (a supplementary figure). In the fovea, the small region of sharpest vision in humans and other higher primates, no inner layers cover the light-detecting cells at all.

Franze et al., “Müller cells are living optical fibers in the vertebrate retina”, PNAS 104, 8287 (2007).

How it was checked. Full text (PubMed Central copy), including the supplementary-figure references quoted above.

Four in five of your neurons sit in the cerebellum

The cerebellum is about a tenth of the brain’s mass and, in the most complete count so far, holds 69 of its 86 billion neurons.

The cerebral cortex, the wrinkled outer layer people picture when they think of a brain, makes up about 82% of brain mass but holds only 19% of its neurons. The cerebellum, tucked under the back of the brain and best known for its role in movement, is about 10% of the mass and holds 80% of the neurons.

A team in Brazil counted them in 2009 by dissolving brain tissue and counting the cell nuclei left behind, in four adult male brains aged 50 to 71. The average was 86.1 billion neurons, with a spread of ±8.1 billion (one standard deviation) between brains. The paper points out that the familiar figure of 100 billion had never been backed by a whole-brain count. Earlier studies counted single structures, and added together they allowed anything from about 75 to 125 billion.

Those earlier regional counts show the same pattern. The cerebellar count is about four times the neocortical one, although the brains, ages and methods differ and the totals cannot simply be added.

A nerve impulse that reaches a synapse often releases nothing

At a single release site, the chance that an arriving impulse releases any transmitter may be well below one half in a living brain.

A synapse is the gap where one neuron signals to the next. When an electrical impulse reaches the sending side, it triggers the release of chemical transmitter only with some probability. J. Gerard G. Borst defines release probability as the average chance that one active zone, a release site on the sending terminal, releases one or more vesicles of transmitter after an impulse. Measurements in cultured neurons and brain slices put the average as high as 0.5.

In living animals the evidence points lower. The size of synaptic potentials there depends little on recent activity, which suggests that release is much less likely than in the dish. Borst’s 2010 review concludes that impulses trigger release much less often under natural conditions. It lists possible reasons for the gap between dish and animal, among them differences in the age of the tissue, higher spontaneous activity, lower calcium outside the cells and tonic inhibition by transmitters in the surrounding fluid. The review gives no single in vivo value in its abstract.

The probability belongs to one release site. If a connection had several sites that acted independently, the chance that an impulse released nothing anywhere would be smaller. The widget below shows that arithmetic; the number of sites in real connections is not taken from the review.

Borst, “The low synaptic release probability in vivo”, Trends in Neurosciences 33, 259 (2010).

How it was checked. Abstract only. The full text is behind a paywall and could not be opened, so what the review says about numbers of release sites, or about the size of the in vivo shortfall, is not reported here.

Thinking hard barely raises your brain’s energy use

The extra energy the brain spends on responding to a task may be as little as 1% of the energy it already uses.

The brain is about 2% of the body’s weight and accounts for about 20% of its energy use. That sounds like a price paid for thought. Marcus Raichle’s 2006 summary of the field says otherwise. He reports an estimate that 60% to 80% of the brain’s energy budget supports communication among neurons and their supporting cells, and that the additional burden from the momentary demands of the environment may be as little as 0.5% to 1.0% of the total. The estimate comes from his review with Mark Mintun, which he cites.

Imaging points the same way. Raichle and Gusnard noted in 2002 that local changes in blood flow during most cognitive tasks are often 5% or less, and that attempts to measure whole-brain changes in blood flow and metabolism during intense mental activity had failed to show any change. Raichle’s name for the large share that responses to stimuli do not explain is “dark energy”, in the astronomers’ sense of something present but unaccounted for.

These are estimates about responses to controlled laboratory tasks.

Raichle, “The brain’s dark energy”, Science 314, 1249 (2006); Raichle and Gusnard, “Appraising the brain’s energy budget”, PNAS 99, 10237 (2002); Raichle and Mintun, “Brain work and brain imaging”, Annu. Rev. Neurosci. 29, 449 (2006).

How it was checked. Full text of Raichle 2006 (a two-page perspective) and of Raichle and Gusnard 2002 (a commentary). Abstract only of Raichle and Mintun, the review behind the 0.5% to 1.0% estimate, which is behind a paywall.

A man blind from two strokes walked a cluttered corridor without a collision

A man left clinically blind by two strokes, with no working visual cortex on the scans, walked a corridor of boxes and chairs with no cane and no guide, and never collided with an obstacle.

The patient, known as TN, suffered two strokes in succession that damaged the visual cortex on each side. Brain imaging found no working visual cortex, and tests found no spared part of his visual field. In daily life he walked like a blind man, with a stick, and needed another person to guide him around the laboratory buildings.

In 2008 a team led by Beatrice de Gelder arranged boxes, chairs and other objects at random along a long corridor, took away his stick and let him walk it with no one guiding him. An experimenter followed behind in case he fell. The authors report that he negotiated the course perfectly and never collided with any obstacle, and a video recording accompanies the paper.

The effect is called blindsight: patients with a destroyed primary visual cortex can detect, locate and sometimes discriminate objects they say they cannot see. The authors conclude that pathways outside the primary visual cortex can support sophisticated spatial skills without perceptual awareness, and say it remains to be determined which pathways account for TN’s navigation. They note that echolocation could not be excluded completely, though they consider it extremely unlikely, and that because his fixation could not be controlled they cannot be certain every part of his visual cortex was destroyed or inactive.

This is one man, and the report gives no collision counts, trajectories or repeated trials. It does not say what he was told beforehand.

de Gelder et al., “Intact navigation skills after bilateral loss of striate cortex”, Current Biology 18, R1128 (2008); Cowey, “The blindsight saga”, Experimental Brain Research 200, 3 (2010).

How it was checked. Full text of the two-page de Gelder paper, read from the author’s own PDF by a second reader (Codex) because the publisher blocks automated access; the publisher’s abstract page was read directly. Abstract only of Cowey.

A cat cut off from its own brain can still step

The spinal cord contains circuits that generate the rhythm of walking, and they keep running when the connection to the brain is cut.

After a cat’s spinal cord is cut completely in the lower back, the brain can no longer command the hind legs. Yet if the front legs rest on a platform and the hind legs are placed on a moving treadmill, the hind legs can walk again. In Barbeau and Rossignol’s 1987 study of adult cats, daily treadmill training brought weight-bearing walking after 3 weeks to 3 months, depending on the animal. Rossignol and Frigon’s 2011 review gives 2 to 3 weeks of daily training for the usual course. The rhythm comes from a network in the spinal cord called a central pattern generator, working with feedback from the legs about where they are and what they touch.

The faster result needs a qualifier. In a 2009 experiment by Frigon and colleagues, four cats first had a partial spinal lesion, were trained to walk and recovered for 64 to 80 days. When the cord was then cut completely, the cats could walk with their hind legs within one day. That speed belongs to a cord that had already adapted, not to an untrained one.

The walking was not normal. The experimenter partly made up for the lost balance by holding the cat’s thorax or tail, and by later stages the foot dragged early in each swing and the knee bent less. Even so, step cycle, stance and swing durations resembled those of normal cats at comparable speeds. How far this independence reaches in people is a separate and open question.

Rossignol and Frigon, “Recovery of locomotion after spinal cord injury: some facts and mechanisms”, Annu. Rev. Neurosci. 34, 413 (2011); Frigon et al., “Asymmetric changes in cutaneous reflexes after a partial spinal lesion and retention following spinalization during locomotion in the cat”, J. Neurophysiol. 102, 2667 (2009); Barbeau and Rossignol, “Recovery of locomotion after chronic spinalization in the adult cat”, Brain Res. 412, 84 (1987).

How it was checked. Abstracts of Frigon 2009 and Barbeau and Rossignol 1987 read directly. The full text of the 2011 review was read by a second reader (Codex) from an author-uploaded copy, for the training times and the one-day result.

In many parts of the nervous system, about half the neurons that form are later eliminated

Developing nervous systems overproduce neurons and then remove many of them by a programme that is part of normal development, and in mice some of that removal happens after birth.

R. W. Oppenheim’s 1991 review opens by observing that in many parts of the central and peripheral nervous systems, roughly one half of the neurons go through a final step of regression and death. The dying neurons carry out a suicide programme, programmed cell death. Later reviews treat it as an adaptive process that evolved for its own purposes, from shaping the neural tube to matching neuron numbers to their targets.

The timing is not limited to the embryo. In a 2012 study of mice, Southwell and colleagues found that 40% of the developing cortical interneurons, the cortex’s inhibitory neurons, were eliminated by Bax-dependent cell death during postnatal life. A 2018 study in mice found that the survival of these interneurons depends on the activity of pyramidal cells in a critical window of postnatal development.

These counts come from animals. This page cites no human percentage.

Oppenheim, “Cell death during development of the nervous system”, Annu. Rev. Neurosci. 14, 453 (1991); Buss, Sun and Oppenheim, “Adaptive roles of programmed cell death during nervous system development”, Annu. Rev. Neurosci. 29, 1 (2006); Southwell et al., “Intrinsically determined cell death of developing cortical interneurons”, Nature 491, 109 (2012); Wong et al., “Pyramidal cell regulation of interneuron survival sculpts cortical networks”, Nature 557, 668 (2018).

How it was checked. First page only of Oppenheim 1991 (the publisher’s free preview; the rest is behind a paywall, so the 20% to 80% range this page used to quote is no longer stated). Abstracts of the other three.

Adults who grew up with one hemisphere still show the standard networks in it

Six adults who had one cerebral hemisphere removed or disconnected as children show the usual seven cortical networks in the hemisphere they have.

Hemispherectomy is surgery for severe epilepsy that isolates the affected hemisphere, either by removing it entirely or, in a functional hemispherectomy, by removing part of it and cutting every connection of what remains.

In 2019 Kliemann and colleagues scanned six adults, aged about 20 to 31, who had had this operation as children, between the ages of 3 months and 11 years. Two had a complete removal and four had a functional hemispherectomy. The authors describe them as high-functioning. The team compared their resting brain activity with six matched controls and with a normative sample of 1,482 people.

The analysis used seven large-scale cortical networks. Within the remaining hemisphere, connectivity among regions of the same network was as strong as in the controls. Connectivity between different networks was markedly higher in almost every participant and across all networks. The authors read this as support for a shared set of networks underlying cognition, and as a possible sign of how the brain reorganises after the operation.

Six people is a small sample. The scans were taken at rest, and the authors leave task-based organisation as an open question. The seven networks are cortical, so the study says nothing about every network in the brain.

Kliemann et al., “Intrinsic functional connectivity of the brain in adults with a single cerebral hemisphere”, Cell Reports 29, 2398 (2019).

How it was checked. Full text (PubMed Central copy), including the methods and the participant table.

How each fact was checked

Every figure above comes from the paper named under it. Two readers were used: the author, and for papers that publishers block, a second reader (Codex) that opened author-hosted or open-access copies. “Full text” below means the whole paper was read; “abstract only” means the paper is behind a paywall and only its abstract was read, so nothing beyond the abstract is claimed from it.

• 1 Hearing. Abstract only: Kemp 1978, Kemp 2002, Penner and Zhang 1997.
• 2 Pain. Full text of the introduction of Jin et al. 2018. Lewis and Pochin 1937 not opened.
• 3 Vision. Full text: Franze et al. 2007.
• 4 Neuron count. Full text: Azevedo et al. 2009. Abstract only: Andersen et al. 1992, Pakkenberg and Gundersen 1997.
• 5 Synapses. Abstract only: Borst 2010.
• 6 Energy. Full text: Raichle 2006, Raichle and Gusnard 2002. Abstract only: Raichle and Mintun 2006.
• 7 Blindsight. Full text: de Gelder et al. 2008 (read by the second reader). Abstract only: Cowey 2010.
• 8 Movement. Abstract: Frigon et al. 2009, Barbeau and Rossignol 1987. Full text of Rossignol and Frigon 2011 (read by the second reader).
• 9 Development. First page only: Oppenheim 1991. Abstract only: Buss et al. 2006, Southwell et al. 2012, Wong et al. 2018.
• 10 Hemispheres. Full text: Kliemann et al. 2019.

What is estimated or still open

• Ears. The prevalence of spontaneous emissions depends on how the recording is analysed, and the figures here are from one study of adults.
• Pain. 15 m/s and 1 m/s are typical values; individual fibres range widely, and the 1 m to the spinal cord is a rough estimate. The slider computes nerve travel time only, not when pain is felt.
• Retina. The experiments were mainly on guinea-pig retina, and the authors’ wording is that Müller cells “seem to mediate” image transfer.
• Neurons. The 2009 counts come from four men aged 50 to 71, and the authors say the cerebellum and rest-of-brain counts are minimums. Larger stereological counts of single structures use different brains and methods. Neuron number is not a measure of importance.
• Synapses. The in vivo release probability is an inference from how synaptic responses behave, not a direct count, and no single value is claimed. The number of release sites on a connection in the widget is hypothetical.
• Energy. The 0.5% to 1.0% figure is an estimate that Raichle cites to a review whose full text was not opened. The 5% figure is a local blood-flow change, not a measure of energy.
• Blindsight. One patient. The report has no counts of near misses, and the authors could not exclude echolocation completely.
• Spinal cats. The timings come from a handful of cats in specific laboratories. Hind-leg stepping with support on a treadmill is established in cats; how much carries over to people with a complete injury is not claimed.
• Cell death. “Roughly half” is a general statement from a review whose full text was not opened; 40% is a single measured population in mice. No human percentage is given.
• Half a brain. Six people is a small sample. The study measured network connectivity at rest in seven cortical networks, not everyday abilities.