experiment news article

Every print subscription comes with full digital access

Science News

Falling objects in orbit show einstein was right — again.

An experiment provides the most precise confirmation yet of a key tenet of general relativity

The final results from the orbiting MICROSCOPE experiment (illustrated) provide the most precise confirmation yet that all things fall the same way under the influence of gravity.

© CNES 2015

Share this:

By James R. Riordon

September 14, 2022 at 10:00 am

Gravity doesn’t discriminate. An experiment in orbit has confirmed, with precision a hundred times greater than previous efforts, that everything falls the same way under the influence of gravity.

The finding is the most stringent test yet of the equivalence principle, a key tenet of Einstein’s theory of general relativity. The principle holds to about one part in a thousand trillion , researchers report September 14 in Physical Review Letters .

The idea that gravity affects all things equally might not seem surprising. But the slightest hint otherwise could help explain how general relativity, the foundational theory of gravity, meshes with the standard model of particle physics, the theoretical framework that describes all fundamental particles of matter. General relativity is a classical theory that sees the universe as smooth and continuous, whereas the standard model is a quantum theory involving grainy bits of matter and energy. Combining them into a single theory of everything has been an unfulfilled dream of scientists extending back to Einstein ( SN: 1/12/22 ).

“The equivalence principle is the most important cornerstone of Einstein’s theory of general relativity,” says Sabine Hossenfelder, a physicist with the Frankfurt Institute for Advanced Studies in Germany who was not involved in the study.  “We know [it] eventually has to be altered because it cannot in its present form take into account quantum effects.”

To help search for potential alterations, the MICROSCOPE experiment tracked the motion of nested metal cylinders — a 300-gram titanium outer cylinder and a 402-gram platinum inner one — as they orbited the Earth in near-perfect free fall. Any difference in the effect of gravity on the respective cylinders would cause them to move relative to each other. Small electrical forces applied to bring the cylinders back into alignment would have revealed a potential violation of the equivalence principle.

From April 2016 to October 2018, the cylinders were shielded inside a satellite that protected them from the buffeting of solar winds, the minuscule pressure that sunlight exerts and the residual atmosphere at an orbital altitude of a little over 700 kilometers high.

By performing the experiment in orbit, the researchers could compare the free fall of two different materials for extended periods without the confounding effects of vibrations or of objects nearby that could exert gravitational forces, says Manuel Rodrigues, a MICROSCOPE team member and physicist with the French aerospace lab ONERA in Palaiseau. “One of the lessons learned by MICROSCOPE is … that space is the best way to get an important improvement in the accuracy for this kind of test.”

Over its two-and-a-half-year mission, MICROSCOPE found no sign of cracks in the equivalence principle, the new study reports. The finding builds on a previous interim report from the experiment that found the same thing , but with less precision ( SN: 12/4/17 ).

Some physicists suspect that limits to the equivalence principle may never turn up in experiments, and that Einstein will perpetually be proven right.

Even 100 times greater precision from a follow-up MICROSCOPE 2 mission, tentatively planned for the 2030s, is unlikely to reveal an equivalence principle breakdown, says Clifford Will, a physicist at the University of Florida in Gainesville who is not affiliated with the experiment. “It really is still this basic idea that Einstein taught,” he says. What we see as the force of gravity is actually the curvature of spacetime. “Any body simply moves along the path in Earth’s spacetime,” whether it’s made of dense platinum, lighter titanium or any other material.

But even if physicists never prove Einstein wrong, Hossenfelder says, experiments like MICROSCOPE are still important. “These tests aren’t just about the equivalence principle,” she says. “They implicitly look for all other kinds of deviations, new forces and so on,” that aren’t part of general relativity. “So really it’s a multiple-purpose, high-precision measurement.”

Now that the mission is complete, the MICROSCOPE satellite will slowly spiral out of orbit. “It’s difficult to bet where in 25 years it will fall down,” Rodrigues says. Along with a reference set of platinum cylinders on board, “it’s [a] couple of millions of euros [in] platinum.” Where that precious platinum metal will land is anyone’s guess, but the gravity that pulls it down will tug on the titanium just as hard, to one part in a thousand trillion at least.

More Stories from Science News on Physics

Zigzag walls could help buildings beat the heat

Paper cut physics pinpoints the most hazardous types of paper

Dark matter experiments get a first peek at the ‘neutrino fog’ 

A new element on the periodic table might be within reach 

Can light spark superconductivity? A new study reignites debate

Jurassic Park ’s amber-preserved dino DNA is now inspiring a way to store data 

Advanced nuclear reactors need a different type of uranium. Here’s 4 things to know 

Something weird is happening to Earth’s inner core

Subscribers, enter your e-mail address for full access to the Science News archives and digital editions.

Not a subscriber? Become one now .

Advertisement

Nuclear fusion experiment overcomes two key operating hurdles

Two important barriers to a stable, powerful fusion reaction have been leapt by an experiment in a small tokamak reactor, but we don’t yet know if the technique will work in larger devices

By Matthew Sparkes

24 April 2024

Inside the DIII-D tokamak fusion reactor

Rswilcox (CC BY-SA 4.0)

A nuclear fusion reaction has overcome two key barriers to operating in a “sweet spot” needed for optimal power production: boosting the plasma density and keeping that denser plasma contained. The milestone is yet another stepping stone towards fusion power, although a commercial reactor is still probably years away.

One of the main avenues being explored in efforts to achieve fusion power is using tokamak reactors. These have a doughnut-shaped chamber where plasma hotter than the surface of our sun is contained by vast magnets .

A new kind of experiment at the LHC could unravel quantum reality

The Large Hadron Collider is testing entanglement in a whole new energy range, probing the meaning of quantum theory – and the possibility that an even stranger reality lies beneath

It had been thought that there was a point – known as the Greenwald limit – above which you couldn’t raise the density of the plasma without it escaping the clutches of the magnets, potentially damaging your reactor. But raising density is crucial to increasing output, as experiments have shown that the output of tokamak reactors rises proportionally with the square of the fuel density.

Now, Siye Ding at General Atomics in San Diego, California, and his colleagues have shown that there is a way to raise the plasma density, and proved that it can be stable, by running the DIII-D National Fusion Facility tokamak reactor for 2.2 seconds with an average density that is 20 per cent above the Greenwald limit. While this barrier has been passed before, with less stability and for shorter durations, this experiment crucially also ran with a metric known as H98(y,2) of above 1.

H98(y,2) is a complex blend of measurements and values that shows how well the plasma is contained by the magnets, says Gianluca Sarri at Queen’s University Belfast, with a value of 1.0 or above signifying that plasma is being successfully held in place.

Sign up to our The Daily newsletter

The latest science news delivered to your inbox, every day.

The gold hydrogen rush: Does Earth contain near-limitless clean fuel?

Prospectors around the world are scrambling to find reserves of "gold hydrogen", a naturally occurring fuel that burns without producing carbon dioxide. But how much is really out there and how easy is it to tap into?

“You’re now starting to show some sort of stable operation where you can consistently be in the sweet spot,” says Sarri. “This was done in a small machine. If you take these results and extrapolate it to a larger machine… that is expected to put you in a situation where gain and significant power production can be achieved over a significant amount of time.”

The DIII-D experiment relied on a mix of approaches that aren’t themselves new, says Sarri, but together seem to have created a promising approach. The team used higher density in the core of the doughnut shaped plasma, to increase output, while allowing it to dip at the edges nearest the containment vessel to drop below the Greenwald limit, therefore avoiding any plasma escape. They also puffed deuterium gas into the plasma to calm reactions in specific spots.

DIII-D’s plasma chamber has an outside radius of just 1.6 metres, and isn’t yet know whether the same method would work for ITER, the next-generation tokamak under construction in France, which will have a radius of 6.2 metres and is expected to create plasma as soon as 2025.

UK nuclear fusion reactor sets new world record for energy output

“These plasmas are very complicated,” says Sarri. “A small change in conditions leads to a big change in behaviour. And experimentally it has been more like a trial-and-error sort of approach, where you try many different configurations and basically see which one is best. It’s all about forcing the plasma to do something that is completely against its nature, that it really doesn’t want to do.”

Ding says the experiment bodes well for the future of fusion power. “Many reactor designs require simultaneous high confinement and high density. Experimentally, this is the first time it is realised,” he says. “The next step is expensive, and currently research is going in many different directions. My hope is that this paper will help focus the efforts worldwide.”

The work is another step towards a practical fusion power plant, says Sarri, but nobody should expect to see a commercial reactor in the next five, or even 10, years.

Journal reference

Nature DOI: 10.1038/s41586-024-07313-3

• energy and fuels /
• nuclear fusion technology /
• futurology /
• nuclear energy

Sign up to our weekly newsletter

Receive a weekly dose of discovery in your inbox! We'll also keep you up to date with New Scientist events and special offers.

More from New Scientist

Explore the latest news, articles and features

Future of fusion: How the UK's JET reactor paved the way for ITER

UK spurns European invitation to join ITER nuclear fusion project

World's first tunnel to a magma chamber could unleash unlimited energy

Subscriber-only

Popular articles

Trending New Scientist articles

NASA Demonstrates ‘Ultra-Cool’ Quantum Sensor for First Time in Space

NASA’s Cold Atom Lab, shown where it’s installed aboard the International Space Station, recently demonstrated the use of a tool called an atom interferometer that can precisely measure gravity and other forces — and has many potential applications in space.

Future space missions could use quantum technology to track water on Earth, explore the composition of moons and other planets, or probe mysterious cosmic phenomena.

NASA’s Cold Atom Lab, a first-of-its-kind facility aboard the International Space Station, has taken another step toward revolutionizing how quantum science can be used in space. Members of the science team measured subtle vibrations of the space station with one of the lab’s onboard tools — the first time ultra-cold atoms have been employed to detect changes in the surrounding environment in space.

The study , which appeared in Nature Communications on Aug. 13, also reports the longest demonstration of the wave-like nature of atoms in freefall in space.

The Cold Atom Lab science team made their measurements with a quantum tool called an atom interferometer, which can precisely measure gravity, magnetic fields, and other forces. Scientists and engineers on Earth use this tool to study the fundamental nature of gravity and advance technologies that aid aircraft and ship navigation. (Cell phones, transistors, and GPS are just a few other major technologies based on quantum science but do not involve atom interferometry.)

Physicists have been eager to apply atom interferometry in space because the microgravity there allows longer measurement times and greater instrument sensitivity, but the exquisitely sensitive equipment has been considered too fragile to function for extended periods without hands-on assistance. The Cold Atom Lab, which is operated remotely from Earth, has now shown it’s possible.

“Reaching this milestone was incredibly challenging, and our success was not always a given,” said Jason Williams, the Cold Atom Lab project scientist at NASA’s Jet Propulsion Laboratory in Southern California. “It took dedication and a sense of adventure by the team to make this happen.”

Power of Precision

Space-based sensors that can measure gravity with high precision have a wide range of potential applications. For instance, they could reveal the composition of planets and moons in our solar system, because different materials have different densities that create subtle variations in gravity.

This type of measurement is already being performed by the U.S.-German collaboration GRACE-FO (Gravity Recovery and Climate Experiment Follow-on), which detects slight changes in gravity to track the movement of water and ice on Earth. An atom interferometer could provide additional precision and stability, revealing more detail about surface mass changes.

Precise measurements of gravity could also offer insights into the nature of dark matter and dark energy, two major cosmological mysteries. Dark matter is an invisible substance five times more common in the universe than the “regular” matter that composes planets, stars, and everything else we can see. Dark energy is the name given to the unknown driver of the universe’s accelerating expansion.

“Atom interferometry could also be used to test Einstein’s theory of general relativity in new ways,” said University of Virginia professor Cass Sackett, a Cold Atom Lab principal investigator and co-author of the new study. “This is the basic theory explaining the large-scale structure of our universe, and we know that there are aspects of the theory that we don’t understand correctly. This technology may help us fill in those gaps and give us a more complete picture of the reality we inhabit.”

A Portable Lab

NASA’s Cold Atom Lab studies the quantum nature of atoms, the building blocks of our universe, in a place that is out of this world – the International Space Station. This animated explainer explores what quantum science is and why NASA wants to do it in space.

About the size of a minifridge, the Cold Atom Lab launched to the space station in 2018 with the goal of advancing quantum science by putting a long-term facility in the microgravity environment of low Earth orbit. The lab cools atoms to almost absolute zero, or minus 459 degrees Fahrenheit (minus 273 degrees Celsius). At this temperature, some atoms can form a Bose-Einstein condensate, a state of matter in which all atoms essentially share the same quantum identity. As a result, some of the atoms’ typically microscopic quantum properties become macroscopic, making them easier to study.

Quantum properties include sometimes acting like solid particles and sometimes like waves. Scientists don’t know how these building blocks of all matter can transition between such different physical behaviors, but they’re using quantum technology like what’s available on the Cold Atom Lab to seek answers.

In microgravity, Bose-Einstein condensates can reach colder temperatures and exist for longer, giving scientists more opportunities to study them. The atom interferometer is among several tools in the facility enabling precision measurements by harnessing the quantum nature of atoms.

We've Got Some Space For You

Due to its wave-like behavior, a single atom can simultaneously travel two physically separate paths. If gravity or other forces are acting on those waves, scientists can measure that influence by observing how the waves recombine and interact.

“I expect that space-based atom interferometry will lead to exciting new discoveries and fantastic quantum technologies impacting everyday life, and will transport us into a quantum future,” said Nick Bigelow, a professor at University of Rochester in New York and Cold Atom Lab principal investigator for a consortium of U.S. and German scientists who co-authored the study.

More About the Mission

A division of Caltech in Pasadena, JPL designed and built Cold Atom Lab, which is sponsored by the Biological and Physical Sciences (BPS) division of NASA’s Science Mission Directorate at the agency’s headquarters in Washington. BPS pioneers scientific discovery and enables exploration by using space environments to conduct investigations that are not possible on Earth. Studying biological and physical phenomena under extreme conditions allows researchers to advance the fundamental scientific knowledge required to go farther and stay longer in space, while also benefitting life on Earth. 

To learn more about Cold Atom Lab, visit:

https://coldatomlab.jpl.nasa.gov/

News Media Contact

Calla Cofield

Jet Propulsion Laboratory, Pasadena, Calif.

626-808-2469

[email protected]

Science News

16th-century compass unearthed in Poland may have belonged to Copernicus

By Margherita Bassi published 19 August 24

The 500-year-old compass is a two-pronged metal device that scientists used for precise measurements in cartography, geometry and astronomy.

Human aging accelerates dramatically at age 44 and 60

By Kamal Nahas published 19 August 24

A small study suggests that people in their 40s and 60s undergo sizable physiological changes that may be connected to age-related illnesses.

Catastrophic collision between Milky Way and Andromeda galaxies may not happen after all, new study hints

By Ben Turner published 19 August 24

Astronomers have long predicted that a collision between our galaxy and nearby Andromeda could be inevitable, but new calculations suggest this may be an over exaggeration.

Quantum data beamed alongside 'classical' data in a single fiber-optic connection for the 1st time

By Owen Hughes published 19 August 24

Scientists have transmitted quantum data and conventional internet data through the same fiber-optic channel for the first time.

Arthur's Stone: A 5,700-year-old monument in England linked to the legend of King Arthur

By Jennifer Nalewicki published 19 August 24

The large stone table is considered one of the most recognizable Neolithic monuments in England.

What was the first animal on Earth?

By Katherine Irving published 19 August 24

Using fossil evidence and genetic dating, geneticists and paleontologists have backed strikingly different candidates for the world's first animal.

Tonight's 'Sturgeon Moon' will be the 1st 'blue supermoon' of 2024: Here's how to see it at its biggest and best

By Jamie Carter last updated 19 August 24

The year's first supermoon is also the third full moon in a summer that includes four, making it a 'blue supermoon'. Here's how to see August's full Sturgeon Moon rise.

Elon Musk said he's 'definitely going to be dead' before humans go to Mars — and you probably will be too

By Brad Bergan published 18 August 24

Book excerpt "If it's taken us 18 years just to get ready to do the first people in orbit, we've got to improve our rate of innovation."

Scientists make breasts in a dish to study lactation

By Gat Rauner published 18 August 24

Scientists are making mammary gland organoids with 3D milk ducts that could shed light on lactation and breast cancer development.

3 shipwrecks from 'forgotten battle' of World War II discovered off remote Alaskan island

By Tom Metcalfe published 18 August 24

This is the first time that wrecks from the almost-overlooked conflict in WWII have been studied scientifically.

Space photo of the week: James Webb telescope catches baby stars roaring to life

By Jamie Carter published 18 August 24

A new image of the Serpens Nebula captured by the James Webb Space Telescope shows that when clouds collapse to form stars, all of those stars spin in the same direction.

When did humans start cooking food?

By Ashley Hamer published 18 August 24

When did humans first roast food over a fire? Archaeology and biology can shed light on this mystery.

Last meal of crocodile mummified in ancient Egypt revealed in CT scans 3,000 years later

By Lidija M. Mcknight published 17 August 24

Silky anteater: The tiny, boxing ball of fur

By Lydia Smith published 17 August 24

The smallest species of anteater grows to just 14 inches, including its tail — but it packs a hefty punch when threatened.

These 'living computers' are made from human neurons — and you can rent one for $500 a month

By Jordan Kinard published 17 August 24

In the search for less energy-hungry artificial intelligence, some scientists are exploring living computers.

New ISS images showcase auroras, moon and space station in glorious photos

By Elizabeth Howell published 17 August 24

The moon's glow meets a stunning aurora in a new astronaut image taken from space.

Do opposites really attract in relationships?

By Alice Sun published 17 August 24

When it comes to love, the adage is that opposites attract. But does that really hold?

Dodos were fast and powerful, not slow and inept, definitive preserved specimen suggests

By Tia Ghose published 16 August 24

A new study has cleared up misconceptions about the extinct dodo, identifying the reference specimen for the species and showing they were fast and powerful.

How will the trapped Boeing Starliner astronauts perceive time after 6 months in space?

By Ruth Ogden, Daniel Eduardo Vigo published 16 August 24

The passage of time can slow to a crawl when you're waiting, and waiting, and waiting. How will the Boeing Starliner astronauts feel time move if they have to stay in space for another 6 months?

Auroras headed to the US again in aftermath of gargantuan 'X-class' solar flare

By Brandon Specktor published 16 August 24

Auroras may once again be visible in northern parts of the U.S. this weekend as Earth braces for impact from a powerful coronal mass ejection.

• View Archive

Sign up for the Live Science daily newsletter now

Get the world’s most fascinating discoveries delivered straight to your inbox.

• 2 Human aging accelerates dramatically at age 44 and 60
• 3 Catastrophic collision between Milky Way and Andromeda galaxies may not happen after all, new study hints
• 4 Quantum data beamed alongside 'classical data' in the same fiber-optic connection for the 1st time
• 5 Arthur's Stone: A 5,700-year-old monument in England linked to the legend of King Arthur

August 9, 2024

Experiments Prepare to Test Whether Consciousness Arises from Quantum Weirdness

Researchers wish to probe whether consciousness has a basis in quantum mechanical phenomena

By Hartmut Neven & Christof Koch

nopparit/Getty Images

The brain is a mere piece of furniture in the vastness of the cosmos, subject to the same physical laws as asteroids, electrons or photons. On the surface, its three pounds of neural tissue seem to have little to do with quantum mechanics , the textbook theory that underlies all physical systems, since quantum effects are most pronounced on microscopic scales. Newly proposed experiments, however, promise to bridge this gap between microscopic and macroscopic systems, like the brain, and offer answers to the mystery of consciousness.

Quantum mechanics explains a range of phenomena that cannot be understood using the intuitions formed by everyday experience. Recall the Schrödinger’s cat thought experiment , in which a cat exists in a superposition of states, both dead and alive. In our daily lives there seems to be no such uncertainty—a cat is either dead or alive. But the equations of quantum mechanics tell us that at any moment the world is composed of many such coexisting states, a tension that has long troubled physicists.

Taking the bull by its horns, the cosmologist Roger Penrose in 1989 made the radical suggestion that a conscious moment occurs whenever a superimposed quantum state collapses. The idea that two fundamental scientific mysteries—the origin of consciousness and the collapse of what is called the wave function in quantum mechanics—are related, triggered enormous excitement.

On supporting science journalism

If you're enjoying this article, consider supporting our award-winning journalism by subscribing . By purchasing a subscription you are helping to ensure the future of impactful stories about the discoveries and ideas shaping our world today.

Penrose’s theory can be grounded in the intricacies of quantum computation . Consider a quantum bit, a qubit, the unit of information in quantum information theory that exists in a superposition of a logical 0 with a logical 1. According to Penrose, when this system collapses into either 0 or 1, a flicker of conscious experience is created, described by a single classical bit.

Penrose, together with anesthesiologist Stuart Hameroff, suggested that such collapse takes place in microtubules , tubelike, elongated structural proteins that form part of the cytoskeleton of cells, such as those making up the central nervous system.

These ideas have never been taken up by the scientific community as brains are wet and warm, inimical to the formation of superpositions, at least compared to existing quantum computers that operate at temperatures 10,000 times colder than room temperature to avoid destroying superposition states.

Penrose’s proposal suffers from a flaw when applied to two or more entangled qubits. Measuring one of these entangled qubits instantaneously reveals the state of the other one, no matter how far away. Their states are correlated, but correlation is not causation, and, according to standard quantum mechanics, entanglement cannot be employed to achieve faster-than-light communication. However, per Penrose’s proposal, qubits participating in an entangled state share a conscious experience. When one of them assumes a definite state, we could use this to establish a communication channel capable of transmitting information faster than the speed of light, a violation of special relativity.

In our view, the entanglement of hundreds of qubits, if not thousands or more, is essential to adequately describe the phenomenal richness of any one subjective experience: the colors, motions, textures, smells, sounds, bodily sensations, emotions, thoughts, shards of memories and so on that constitute the feeling of life itself.

In an article published in the open-access journal Entropy , we and our colleagues turned the Penrose hypothesis on its head, suggesting that an experience is created whenever a system goes into a quantum superposition rather than when it collapses. According to our proposal, any system entering a state with one or more entangled superimposed qubits will experience a moment of consciousness.

You, the astute reader, must by now be saying to yourself: But wait a minute here—I don’t ever consciously experience a superposition of states. Any one experience has a definitive quality; it is one thing and not the other. I see a particular shade of red, feel a toothache. I don’t simultaneously experience red and not-red, pain and not-pain.

The definitiveness of any conscious experience naturally arises within the many-worlds interpretation of quantum mechanics . A metaphysical position first put forward by physicist Hugh Everett in 1957, the many-worlds view, posits time’s evolution as an enormously branched tree, with every possible outcome of a quantum event splitting off its own universe. A single qubit entering a superposition gives birth to two universes, in one of which the qubit’s state is 0 while in a twin universe everything is identical except that the qubit’s state is 1.

Entanglement potentially offers something else for brain scientists by providing a natural solution to what is called the binding problem, the subjective unity of every experience that has long posed a key challenge to the study of consciousness. Consider seeing the Statue of Liberty: her face, the crown on her head, the torch in her raised right hand, and so on. All these distinctions and relationships are bound together into a single perception whose substrate might be numerous qubits, all entangled with each other.

To make these esoteric ideas concrete, we propose three experiments that would increasingly shape our thinking on these matters. The first experiment, progressing right now thanks to funding from the Santa Monica–based Tiny Blue Dot Foundation, seeks to provide evidence of the relevance of quantum mechanics to neuroscience in two very accessible test beds: tiny fruit flies and cerebral organoids, the latter lentil-sized assemblies of thousands of neurons grown from human-induced pluripotent stem cells. It is known that the inert noble gas xenon can act as anesthetic in animals and people. Remarkably, an earlier experiment claimed that its anesthetic potency, measured as the concentration of the gas that induces immobility, depends on the specific isotopes of xenon. Two isotopes of an element contain the same number of positively charged protons but different numbers of noncharged neutrons in their nuclei. The chemical properties of isotopes—that is, what they interact with—are similar, by and large, even though their masses and magnetic properties differ slightly.

If fruit flies and organoids can be used to detect different xenon isotopes, the hunt will be on for the exact mechanisms by which a gas that is inert and that remains aloof from binding to proteins or other molecules achieves this. Is it the tiny difference in the mass of these isotopes (131 versus 132 nucleons) that makes the difference? Or is it their nuclear spin, a quantum mechanical property of the nucleus? These xenon isotopes differ substantially in their nuclear spin; some have zero spin and others 1 / 2 or 3 / 2 .

These xenon experiments will inform a second follow-on experiment in which we will attempt to couple qubits to brain organoids in a way that allows entanglement to spread between biological and technical qubits. The final experiment, which at this stage is still a purely conceptual one, aims to enhance consciousness by coupling engineered quantum states to a human brain in an entangled manner. The person may then experience an expanded state of consciousness like those accessed under the influence of ayahuasca or psilocybin.

Both quantum engineering and the design of brain-machine interfaces are progressing rapidly. It may not be beyond human ingenuity to directly probe and expand our conscious mind by making use of quantum science and technology.

This is an opinion and analysis article, and the views expressed by the author or authors are not necessarily those of Scientific American.

The same toxin targets different receptors in prey and conspecifics to deliver either a lethal or non-lethal blow.

experiments

Browsing: Physics

Read interesting physics news and the latest physics research discoveries on SciTechDaily. Your premier source for the latest revelations, innovations, and research in the captivating world of physics includes recent breakthroughs from sources like Harvard , MIT , Los Alamos , Rice University , Princeton , and Lawrence Berkeley .

We bring you up-to-the-minute information on a wide array of topics, spanning from fundamental physics and quantum mechanics to fluid dynamics, particle physics, and beyond. Our expertly curated content explores the diverse aspects of the universe, unveiling the underlying principles that govern its behavior and uncovering the mysteries that continue to intrigue scientists and enthusiasts alike. Stay informed about groundbreaking discoveries, technological advancements, and theoretical breakthroughs that deepen our understanding of the cosmos and reshape our perspective on reality.

Popular physics news topics include Particle , Nuclear , and Quantum Physics , as well as Astrophysics , Biophysics , Heliophysics , Geophysics , and Quantum Computing .

Redefining Quantum Limits: Physicists Unlock the Secret of Elusive Negative Entanglement Entropy

Researchers have successfully demonstrated negative entanglement entropy using classical electrical circuits as stand-ins for complex…

DUNE Milestone Achieved: Completion of Massive Caverns Deep Underground for Neutrino Research

The Sanford Underground Research Facility celebrated the completion of excavation for the LBNF/DUNE project, a…

Large Hadron Collider’s Beam Pipe Edges Closer to Unmasking Magnetic Monopoles

Researchers used a section of the LHC’s beam pipe to hunt for magnetic monopoles, hypothetical…

Physicists Throw World’s Tiniest Disco Party With Levitated Nanodiamonds

A new milestone in levitated optomechanics has been achieved by Prof. Tongcang Li’s group, who…

Electrons Defy Expectations: Quantum Discoveries Unveil New States of Matter

Researchers have uncovered new phenomena in the study of fractional quantum Hall effects. Their experiments,…

How Sonic Technology Is Advancing Wind Detection on Mars

Mars’s hostile environment presents significant challenges for measuring wind speeds, but new advancements are making…

Breaking Barriers in Optics: The Power of Coherence Entropy

Scientists have made a groundbreaking advancement in understanding light propagation through complex media, potentially revolutionizing…

Two Mile Long X-Ray Laser Opens New Windows Into a Mysterious State of Matter

Physicists have made significant strides in understanding warm dense matter using the European XFEL, observing…

Breaking Barriers: First-Ever Quantum Memory for X-Rays Achieved

Researchers have developed a groundbreaking quantum memory for X-rays, allowing extended memory times and paving…

Shattering Big Bang Myths: Surprising Insights Into the Origins of Matter in the Early Universe

Scientists have recreated the extreme conditions of the early universe in particle accelerators, revealing surprising…

Physicists Develop Next-Gen Optical Fibers for Future Quantum Data Transfer

Physicists at the University of Bath in the UK have developed a new generation of…

A Simpler Path to Fusion: The Promise of Spherical Tokamak Technology

Carving a new path forward for compact fusion vessels. Some experts believe that the future…

From Chaos to Choreography: Revealing the Hidden Laws of Collective Motion

Physical and biological systems are different. But are they? A new study on JSTAT observes…

Electron Dynamics Redefined Through Super-Bloch Oscillations

Researchers achieve advances in periodic oscillations and transportation for optical pulses, with potential for next-gen…

Matter vs. Antimatter: New CERN Experiment Challenges Conventional Physics

The BASE team at CERN has enhanced antiproton cooling techniques, allowing quicker and more precise…

Harnessing the Power of Micro-Bunching: A New Frontier in Synchrotron Radiation

A new technique shortens electron bunches in storage rings to produce more powerful, coherent synchrotron…

Unlocking Future Technologies With Magnetic Control of Rare Earth Elements

Laser pulses have been shown to adjust the magnetic properties of rare earths by affecting…

Astonishing Nuclear Breakthrough Could Rewrite the Fundamental Constants of Nature

The findings may lead to the creation of the most precise clock ever, facilitating advancements…

Type above and press Enter to search. Press Esc to cancel.

It’s a wonderful world — and universe — out there.

Come explore with us!  

Science News Explores

Experiments.

This series is meant to teach about how science is done, from generating a hypothesis to designing an experiment to analyzing the results with statistics. You can repeat the steps of one of these experiments and compare your results — or use this as inspiration to design your own experiment.

Experiment: Why does the moon look larger on the horizon?

Experiment: Measure the speed of light with a laser

Experiment: How to make the boldest, brightest tie-dye!

Experiment: Make your own cents-able battery

Experiment: Can plants stop soil erosion?

Experiment: Are we there yet? Test how migratory birds navigate

Experiment: Where does a bouncing basketball’s energy go?

Experiment: How well do different materials create static electricity?

Educators and parents, sign up for the cheat sheet.

Weekly updates to help you use Science News Explores in the learning environment

Thank you for signing up!

There was a problem signing you up.

Experiment: Kimchi chemistry

Take candy core samples with this science activity

Experiment: A puzzling parallax helps stargazers

Experiment: Test the effect of temperature on reaction time

• Skip to main content
• Keyboard shortcuts for audio player

Goats and Soda

• Infectious Disease
• Development
• Women & Girls
• Coronavirus FAQ

It's one of the biggest experiments in fighting global poverty. Now the results are in

Nurith Aizenman

Husband and wife Denis and Bentha Otieno at their home in 2017, calculating their monthly budget shortly after they began receiving a monthly grant from the charity GiveDirectly. Researchers are studying whether the grant program — which provides the equivalent of $50 every month over 12 years — can lift people out of poverty. Nichole Sobecki for NPR hide caption

It's an unprecedented – and massive – experiment: Since 2017 the U.S.-based charity GiveDirectly has been providing thousands of villagers in Kenya what's called a "universal basic income" – a cash grant that's the equivalent of about $50, delivered every month, with the commitment to keep the payments coming for 12 years. It is a crucial test of what many consider one of the most cutting-edge ideas for alleviating global poverty. This week a team of independent researchers who have been studying the impact released their first results .

Their findings cover the first two years of the effort and compare the outcomes for about 5,000 people who got the monthly payments to nearly 12,000 others in a control group who got no money. But, just as significantly, the researchers also compared the recipients to people in two other categories: nearly 9,000 who received the monthly income for just two years, without the promise of another decade of payments afterward; and another roughly 9,000 people who got that same two years' worth of income but in a lump-sum payment.

NPR has been covering this effort from the start — traveling to Kenya early into the launch at a village near Lake Victoria. During a community meeting that day people's phones suddenly began to ping with a text alert, notifying them that their monthly grant had just been sent to their mobile bank accounts. The crowd erupted in cheers. Some of the younger women broke into song. The joy was a reflection of just how much people in the community had been struggling: The year before this experiment started, 85% of recipients reported experiencing hunger.

So how much of a difference has the experiment made so far? Here are five takeaways from the first batch of findings:

1. Giving cash aid in a lump sum has some major advantages over parceling it out.

When it came to measures of well-being such as consumption of protein or spending money on schooling, all of the groups who were given cash were better off than people in the control group who got no money. This fits with previous studies of no-strings cash aid, which find that poor people generally use the money productively rather than wasting it on alcohol, cigarettes or other vices.

But the big news came on a different measure: people's likelihood of starting a business. On this front, those who got the money in a lump sum vastly outperformed people who were promised the same amount for just two years but received it in monthly installments. For instance lump-sum recipients had 19% more enterprises – businesses such as small shops in local markets, motorbike taxis and small-scale construction concerns. And the lump sum recipients' net revenues from their businesses were a whopping 80% higher.

A member of the research team, MIT economist Tavneet Suri , says these results add to the evidence that many poor people are trapped in poverty by a lack of capital for precisely the kinds of transformative investments they would need to vault them into higher incomes.

"I might have this amazing opportunity to invest that's going to get me great returns," says Suri. "But there's no way to borrow. I don't have title to my land, so I can't use my land as collateral. Or I just don't have great ways to save money – because putting it under my mattress is not a great way to save." In short, without an intervention like the lump-sum grants, she says, an individual struggling with poverty might think, "I can't make this investment that would help get me out of poverty."

2. Lump sums are so useful that even those who didn't get them have banded together to create their own version.

GiveDirectly 's head of research, Miriam Laker-Oketta, notes that it wasn't all that surprising that the study team, which worked independently of her organization, found that the lump-sum recipients were more likely to make investments compared with those who got paid in monthly installments. Prior studies of smaller scale cash-aid programs — including an earlier experiment arranged by GiveDirectly itself — have pointed to similar results.

But this new experiment tests, for the first time, both the lump sums and the two years worth of monthly installments against the much larger promise of 12 years of income, again delivered in monthly installments.

So it's notable that here too, the lump-sum recipients did best in the matchup – opening more businesses and earning more money from them even when compared to those who knew they'd be getting monthly payments for the full 12 years.

Andrew Zeitlen, an economist at Georgetown University who studies cash aid, says it's an impressive finding of a "well-executed study." After all, says Zeitlin, who was not involved with the research, "the long-run value of that universal basic income substantially exceeds the value of the lump sum transfers. It's an order of magnitude difference." So, the fact that lump sums had more impact even than this much bigger eventual payout points to the advantage of giving money at once instead of piecemeal.

A 2017 meeting of a rotating savings club formed in a village near Lake Victoria soon after every adult there was chosen to receive a monthly through GiveDirectly's experiment. The clubs have enabled recipients to convert their grants into lump sum payments: Each month the members put $10 into the communual pot — for a total of $100 — and a different person takes it home. Nichole Sobecki for NPR hide caption

A 2017 meeting of a rotating savings club formed in a village near Lake Victoria soon after every adult there was chosen to receive a monthly through GiveDirectly's experiment. The clubs have enabled recipients to convert their grants into lump sum payments: Each month the members put $10 into the communual pot — for a total of $100 — and a different person takes it home.

Just as important, says Suri, is a second twist: Those who were promised 12 years of monthly payments still out-performed people who could only count on two years of payments. And – here's the key – the way that the 12-year-group was able to invest more in their enterprises was by effectively converting their monthly payments into a lump sum.

They did this by making use of a creative financing tool known as a "rotating savings club." Every month members of the club pool their money and then take turns getting the entire payout from that pot.

Rotating savings clubs are enormously popular among Kenyans who don't have access to traditional banking. Even people who got the monthly income for just two years managed to put about 8% more money in a rotating savings club than those who got no aid.

But people in the 12-year-monthly income group used the clubs at an astonishing rate – contributing nearly 70% more money than those in the control group.

Suri says one explanation could be that people who were promised a full 12 years of monthly income knew their neighbors would also be getting the income because every adult in the village was made that same promise. This expectation of years of income to come for everyone involved likely provided people the confidence needed to invest in a savings club: After all, says Suri, you're relying on your fellow members to keep contributing to the pot after they've gotten their own payout.

A meeting of another rotating savings club in the same village, this one founded by Denis Otieno (third from the right). People who were promised the monthly income grants for 12 years used such clubs at an astonishing rate – contributing nearly 70% more money than those in the control group that got no aid. Nichole Sobecki for NPR hide caption

3. Making the benefit 'universal' – by paying every adult in the village – seems to have greatly increased the impact.

This broad-based, "universal" nature of the aid may also help explain another surprising finding, says Suri: The fact that people who chose to invest their cash grants did so by starting businesses.

"I thought we would see tons of investment in agriculture" – basically improvements to the tiny plots on which many villagers raise subsistence crops, she says. "Go buy fertilizer. Go buy a pump to bring in more water."

That's what earlier studies suggested.

But the prior interventions that those studies had analyzed were not "universal" in the sense that, instead, the aid was given to only a subset of people in a community. By contrast, this experiment – by providing the aid to every adult in a given village – "allows us to learn about the interdependence between people," says Zeitlin. In particular, he says, it shows how the aid could boost businesses not just with capital but also by creating a large pool of new potential customers.

Suri says anecdotal evidence suggests this is precisely what happened. "It's everybody getting the aid, and everybody knows that," she says.

4. The grants did not seem to fuel inflation

Despite the sudden influx of money into these impoverished communities, Suri says that so far the data suggests that inflation there did not go up.

One possible reason, she says, is that while people did buy more things, this extra spending was distributed over a wide range of products, depending on the relative wealth of the person getting the aid.

"So it's not all going into one commodity," says Suri. "And that's the advantage of spreading it universally."

5. The big remaining question is whether the benefits of lump-sum payments actually last.

Suri says the findings thus far already have potential implications for policy. For instance, at present, "a lot of cash transfers that the World Bank runs in poor countries tend to be of the monthly-for-two-years kind of style." And this new data adds substantial evidence to the view that, in fact, "the short-term [parceled out aid] is probably not such a smart policy. Because you could take the money and give it in a lump sum and get much bigger effects."

What remains to be seen, she says, is whether the relative benefits of the lump-sum payments endure. Are the businesses that people start durable? Do they generate enough income to actually lift people out of poverty?

"The lump sum and the long term [monthly payments] look similar at two years," Suri says. "But the question is, does the lump sum [impact] fade after year five? Year six? Does it just disappear? Or was this enough to keep [the impacts] going forever?"

Because if so, she adds, "Then we're good. I don't have to spend 12 years of money. I just have to spend two years' worth and just structure it correctly."

To find the answers, Suri says she's committed to continuing this study for as long as it takes.

"For the rest of my life," she says, laughing. "You know, most people want to write a will for their assets – like, who are they going to leave their money to? I'm like, 'Who am I going to leave the universal basic income project to?' It's maybe the most valuable thing I have as a researcher."

• Universal Basic Income
• GiveDirectly

Articles on Experimental psychology

Displaying all articles.

Does competition make us less moral? New research says yes, but only a little bit

Ozan Isler , The University of Queensland

6 eyewitnesses misidentified a murderer – here’s what went wrong in the lineup

Laura Smalarz , Arizona State University

Women preferred for STEM professorships – as long as they’re equal to or better than male candidates

Stephen J Ceci , Cornell University and Wendy M Williams , Cornell University

Recreating language’s Big Bang through a game of vocal charades

Marcus Perlman , University of Wisconsin-Madison

Children with gender dysphoria deserve better from the law

Malcolm Smith , Queensland University of Technology

Virtual bodyswapping reduces bias against other races

Manos Tsakiris , Royal Holloway University of London

Anxiety makes us more biased to threat but exercise can change our perception

Adam Heenan , Queen's University, Ontario

Debt collectors are the true ancestors of Facebook’s emotions experiment

Joe Deville , Goldsmiths, University of London

Revisiting Milgram’s shocking obedience experiments

Nick Haslam , The University of Melbourne and Gina Perry , The University of Melbourne

Women act like men when they switch seats at speed dating

Gary W. Lewandowski Jr. , Monmouth University

Cheaters use cognitive tricks to feel good about themselves

Benjamin Le , Haverford College

Ethics stop dodgy experiments but there’s too much red tape

Dr Gerasimos Chatzidamianos , Lancaster University

Related Topics

• Academic hiring
• Cognitive science
• Linguistics
• Social networks

Top contributors

Professor of Human Development, Cornell University

Professor of Psychology, Monmouth University

Senior Lecturer in Psychology, Manchester Metropolitan University

PhD Candidate in Clinical Psychology, Queen's University, Ontario

Research Fellow, School of Economics, The University of Queensland

Assistant Professor of Psychology, Arizona State University

Senior Lecturer, Australian Centre for Health Law Research, Queensland University of Technology

Postdoctoral Research Associate in Psychology, University of Wisconsin-Madison

Associate Professor of Psychology & Department Chair, Haverford College

Lecturer in Mobile Work, Lancaster University

Science historian with a specific interest in the history of social psychology., The University of Melbourne

Professor of Psychology, Director of the Centre for the Politics of Feelings, Royal Holloway University of London

Professor of Psychology, The University of Melbourne

• X (Twitter)
• Unfollow topic Follow topic

Suggestions or feedback?

MIT News | Massachusetts Institute of Technology

• Machine learning
• Sustainability
• Black holes
• Classes and programs

Departments

• Aeronautics and Astronautics
• Brain and Cognitive Sciences
• Architecture
• Political Science
• Mechanical Engineering

Centers, Labs, & Programs

• Abdul Latif Jameel Poverty Action Lab (J-PAL)
• Picower Institute for Learning and Memory
• Lincoln Laboratory
• School of Architecture + Planning
• School of Engineering
• School of Humanities, Arts, and Social Sciences
• Sloan School of Management
• School of Science
• MIT Schwarzman College of Computing

LLMs develop their own understanding of reality as their language abilities improve

Press contact :.

Previous image Next image

Ask a large language model (LLM) like GPT-4 to smell a rain-soaked campsite, and it’ll politely decline. Ask the same system to describe that scent to you, and it’ll wax poetic about “an air thick with anticipation" and “a scent that is both fresh and earthy," despite having neither prior experience with rain nor a nose to help it make such observations. One possible explanation for this phenomenon is that the LLM is simply mimicking the text present in its vast training data, rather than working with any real understanding of rain or smell. But does the lack of eyes mean that language models can’t ever “understand" that a lion is “larger" than a house cat? Philosophers and scientists alike have long considered the ability to assign meaning to language a hallmark of human intelligence — and pondered what essential ingredients enable us to do so. Peering into this enigma, researchers from MIT’s Computer Science and Artificial Intelligence Laboratory (CSAIL) have uncovered intriguing results suggesting that language models may develop their own understanding of reality as a way to improve their generative abilities. The team first developed a set of small Karel puzzles, which consisted of coming up with instructions to control a robot in a simulated environment. They then trained an LLM on the solutions, but without demonstrating how the solutions actually worked. Finally, using a machine learning technique called “probing,” they looked inside the model’s “thought process” as it generates new solutions.  After training on over 1 million random puzzles, they found that the model spontaneously developed its own conception of the underlying simulation, despite never being exposed to this reality during training. Such findings call into question our intuitions about what types of information are necessary for learning linguistic meaning — and whether LLMs may someday understand language at a deeper level than they do today. “At the start of these experiments, the language model generated random instructions that didn’t work. By the time we completed training, our language model generated correct instructions at a rate of 92.4 percent,” says MIT electrical engineering and computer science (EECS) PhD student and CSAIL affiliate Charles Jin, who is the lead author of a new paper on the work . “This was a very exciting moment for us because we thought that if your language model could complete a task with that level of accuracy, we might expect it to understand the meanings within the language as well. This gave us a starting point to explore whether LLMs do in fact understand text, and now we see that they’re capable of much more than just blindly stitching words together.” Inside the mind of an LLM

The probe helped Jin witness this progress firsthand. Its role was to interpret what the LLM thought the instructions meant, unveiling that the LLM developed its own internal simulation of how the robot moves in response to each instruction. As the model’s ability to solve puzzles improved, these conceptions also became more accurate, indicating that the LLM was starting to understand the instructions. Before long, the model was consistently putting the pieces together correctly to form working instructions. Jin notes that the LLM’s understanding of language develops in phases, much like how a child learns speech in multiple steps. Starting off, it’s like a baby babbling: repetitive and mostly unintelligible. Then, the language model acquires syntax, or the rules of the language. This enables it to generate instructions that might look like genuine solutions, but they still don’t work.

The LLM’s instructions gradually improve, though. Once the model acquires meaning, it starts to churn out instructions that correctly implement the requested specifications, like a child forming coherent sentences. Separating the method from the model: A “Bizarro World”

The probe was only intended to “go inside the brain of an LLM” as Jin characterizes it, but there was a remote possibility that it also did some of the thinking for the model. The researchers wanted to ensure that their model understood the instructions independently of the probe, instead of the probe inferring the robot’s movements from the LLM’s grasp of syntax.

“Imagine you have a pile of data that encodes the LM’s thought process,” suggests Jin. “The probe is like a forensics analyst: You hand this pile of data to the analyst and say, ‘Here’s how the robot moves, now try and find the robot’s movements in the pile of data.’ The analyst later tells you that they know what’s going on with the robot in the pile of data. But what if the pile of data actually just encodes the raw instructions, and the analyst has figured out some clever way to extract the instructions and follow them accordingly? Then the language model hasn't really learned what the instructions mean at all.”

To disentangle their roles, the researchers flipped the meanings of the instructions for a new probe. In this “Bizarro World,” as Jin calls it, directions like “up” now meant “down” within the instructions moving the robot across its grid.  “If the probe is translating instructions to robot positions, it should be able to translate the instructions according to the bizarro meanings equally well,” says Jin. “But if the probe is actually finding encodings of the original robot movements in the language model’s thought process, then it should struggle to extract the bizarro robot movements from the original thought process.” As it turned out, the new probe experienced translation errors, unable to interpret a language model that had different meanings of the instructions. This meant the original semantics were embedded within the language model, indicating that the LLM understood what instructions were needed independently of the original probing classifier. “This research directly targets a central question in modern artificial intelligence: are the surprising capabilities of large language models due simply to statistical correlations at scale, or do large language models develop a meaningful understanding of the reality that they are asked to work with? This research indicates that the LLM develops an internal model of the simulated reality, even though it was never trained to develop this model,” says Martin Rinard, an MIT professor in EECS, CSAIL member, and senior author on the paper.

This experiment further supported the team’s analysis that language models can develop a deeper understanding of language. Still, Jin acknowledges a few limitations to their paper: They used a very simple programming language and a relatively small model to glean their insights. In an  upcoming work , they’ll look to use a more general setting. While Jin’s latest research doesn’t outline how to make the language model learn meaning faster, he believes future work can build on these insights to improve how language models are trained.

“An intriguing open question is whether the LLM is actually using its internal model of reality to reason about that reality as it solves the robot navigation problem,” says Rinard. “While our results are consistent with the LLM using the model in this way, our experiments are not designed to answer this next question.”

“There is a lot of debate these days about whether LLMs are actually ‘understanding’ language or rather if their success can be attributed to what is essentially tricks and heuristics that come from slurping up large volumes of text,” says Ellie Pavlick, assistant professor of computer science and linguistics at Brown University, who was not involved in the paper. “These questions lie at the heart of how we build AI and what we expect to be inherent possibilities or limitations of our technology. This is a nice paper that looks at this question in a controlled way — the authors exploit the fact that computer code, like natural language, has both syntax and semantics, but unlike natural language, the semantics can be directly observed and manipulated for experimental purposes. The experimental design is elegant, and their findings are optimistic, suggesting that maybe LLMs can learn something deeper about what language ‘means.’”

Jin and Rinard’s paper was supported, in part, by grants from the U.S. Defense Advanced Research Projects Agency (DARPA). 

Share this news article on:

Related links.

• Charles Jin
• Martin Rinard
• Computer Science and Artificial Intelligence Laboratory (CSAIL)
• Department of Electrical Engineering and Computer Science

Related Topics

• Artificial intelligence
• Programming
• Computer science and technology
• Electrical Engineering & Computer Science (eecs)
• Defense Advanced Research Projects Agency (DARPA)

Related Articles

Reasoning skills of large language models are often overestimated

Understanding the visual knowledge of language models

Natural language boosts LLM performance in coding, planning, and robotics

Exact symbolic artificial intelligence for faster, better assessment of AI fairness

Previous item Next item

More MIT News

AI assistant monitors teamwork to promote effective collaboration

Read full story →

MIT study explains why laws are written in an incomprehensible style

More durable metals for fusion power reactors

3 Questions: How to prove humanity online

When the lights turned on in the universe

Lincoln Laboratory and National Strategic Research Institute launch student research program to tackle biothreats to national security

• More news on MIT News homepage →

Massachusetts Institute of Technology 77 Massachusetts Avenue, Cambridge, MA, USA

• Map (opens in new window)
• Events (opens in new window)
• People (opens in new window)
• Careers (opens in new window)
• Accessibility
• Social Media Hub
• MIT on Facebook
• MIT on YouTube
• MIT on Instagram

Thank you for visiting nature.com. You are using a browser version with limited support for CSS. To obtain the best experience, we recommend you use a more up to date browser (or turn off compatibility mode in Internet Explorer). In the meantime, to ensure continued support, we are displaying the site without styles and JavaScript.

• View all journals
• Explore content
• About the journal
• Publish with us
• Sign up for alerts
• Published: 18 February 2016

Modern Milgram experiment sheds light on power of authority

• Alison Abbott  

Nature volume  530 ,  pages 394–395 ( 2016 ) Cite this article

11k Accesses

2 Citations

417 Altmetric

Metrics details

• Neuroscience

People obeying commands feel less responsibility for their actions.

More than 50 years after a controversial psychologist shocked the world with studies that revealed people’s willingness to harm others on order, a team of cognitive scientists has carried out an updated version of the iconic ‘Milgram experiments’.

Their findings may offer some explanation for Stanley Milgram's uncomfortable revelations: when following commands, they say, people genuinely feel less responsibility for their actions — whether they are told to do something evil or benign.

“If others can replicate this, then it is giving us a big message,” says neuroethicist Walter Sinnot-Armstrong of Duke University in Durham, North Carolina, who was not involved in the work. “It may be the beginning of an insight into why people can harm others if coerced: they don’t see it as their own action.”

The study may feed into a long-running legal debate about the balance of personal responsibility between someone acting under instruction and their instructor, says Patrick Haggard, a cognitive neuroscientist at University College London, who led the work, published on 18 February in Current Biology 1 .

Milgram’s original experiments were motivated by the trial of Nazi Adolf Eichmann, who famously argued that he was ‘just following orders’ when he sent Jews to their deaths. The new findings don’t legitimize harmful actions, Haggard emphasizes, but they do suggest that the ‘only obeying orders’ excuse betrays a deeper truth about how a person feels when acting under command.

Ordered to shock

In a series of experiments at Yale University in New Haven, Connecticut, in the 1960s, Milgram told his participants that a man was being trained to learn word pairs in a neighbouring room. The participants had to press a button to deliver an electric shock of escalating strength to the learner when he made an error; when they did so, they heard his cries of pain. In reality, the learner was an actor, and no shock was ever delivered. Milgram’s aim was to see how far people would go when they were ordered to step up the voltage.

Routinely, an alarming two-thirds of participants continued to step up shocks, even after the learner was apparently rendered unconscious. But Milgram did not assess his participants’ subjective feelings as they were coerced into doing something unpleasant. And his experiments have been criticized for the deception that they involved — not just because participants may have been traumatized, but also because some may have guessed that the pain wasn’t real.

Modern teams have conducted partial and less ethically complicated replications of Milgram’s work. But Haggard and his colleagues wanted to find out what participants were feeling. They designed a study in which volunteers knowingly inflicted real pain on each other, and were completely aware of the experiment’s aims.

Because Milgram’s experiments were so controversial, Haggard says that he took “quite a deep breath before deciding to do the study”. But he says that the question of who bears personal responsibility is so important to the rule of law that he thought it was “worth trying to do some good experiments to get to the heart of the matter.”

Sense of agency

In his experiments, the volunteers (all were female, as were the experimenters, to avoid gender effects) were given £20 (US$29). In pairs, they sat facing each other across a table, with a keyboard between them. A participant designated the ‘agent’ could press one of two keys; one did nothing. But for some pairs, the other key would transfer 5p to the agent from the other participant, designated the ‘victim’; for others, the key would also deliver a painful but bearable electric shock to the victim’s arm. (Because people have different tolerances to pain, the level of the electric shock was determined for each individual before the experiment began.) In one experiment, an experimenter stood next to the agent and told her which key to press. In another, the experimenter looked away and gave the agent a free choice about which key to press.

To examine the participants’ ‘sense of agency’ — the unconscious feeling that they were in control of their own actions — Haggard and his colleagues designed the experiment so that pressing either key caused a tone to sound after a few hundred milliseconds, and both volunteers were asked to judge the length of this interval. Psychologists have established that people perceive the interval between an action and its outcome as shorter when they carry out an intentional action of their own free will, such as moving their arm, than when the action is passive, such as having their arm moved by someone else.

When they were ordered to press a key, the participants seemed to judge their action as more passive than when they had free choice — they perceived the time to the tone as longer.

In a separate experiment, volunteers followed similar protocols while electrodes on their heads recorded their neural activity through EEG (electroencephalography). When ordered to press a key, their EEG recordings were quieter — suggesting, says Haggard, that their brains were not processing the outcome of their action. Some participants later reported feeling reduced responsibility for their action.

Unexpectedly, giving the order to press the key was enough to cause the effects, even when the keystroke led to no physical or financial harm. “It seems like your sense of responsibility is reduced whenever someone orders you to do something — whatever it is they are telling you to do,” says Haggard.

The study might inform legal debate, but it also has wider relevance to other domains of society, says Sinnot-Armstrong. For example, companies that want to create — or avoid — a feeling of personal responsibility among their employees could take its lessons on board.

Caspar, E. A., Christensen, J. F., Cleeremans, A. & Haggard, P. Curr. Biol. http://dx.doi.org/10.1016/j.cub.2015.12.067 (2016).

Download references

You can also search for this author in PubMed   Google Scholar

Additional information

Tweet Facebook LinkedIn Weibo

Related links

Related links in nature research.

Experimental psychology: The anatomy of obedience 2015-Jul-22

Virtual reality shocker 2006-Dec-22

Related external links

UCL Institute of Cognitive Neuroscience

Rights and permissions

Reprints and permissions

About this article

Cite this article.

Abbott, A. Modern Milgram experiment sheds light on power of authority. Nature 530 , 394–395 (2016). https://doi.org/10.1038/nature.2016.19408

Download citation

Published : 18 February 2016

Issue Date : 25 February 2016

DOI : https://doi.org/10.1038/nature.2016.19408

Share this article

Anyone you share the following link with will be able to read this content:

Sorry, a shareable link is not currently available for this article.

Provided by the Springer Nature SharedIt content-sharing initiative

Quick links

• Explore articles by subject
• Guide to authors
• Editorial policies

Sign up for the Nature Briefing newsletter — what matters in science, free to your inbox daily.

• - Google Chrome

Intended for healthcare professionals

• My email alerts
• BMA member login
• Username * Password * Forgot your log in details? Need to activate BMA Member Log In Log in via OpenAthens Log in via your institution

Search form

• Advanced search
• Search responses
• Search blogs
• Beauty sleep:...

Beauty sleep: experimental study on the perceived health and attractiveness of sleep deprived people

• Related content
• Peer review
• John Axelsson , researcher 1 2 ,
• Tina Sundelin , research assistant and MSc student 2 ,
• Michael Ingre , statistician and PhD student 3 ,
• Eus J W Van Someren , researcher 4 ,
• Andreas Olsson , researcher 2 ,
• Mats Lekander , researcher 1 3
• 1 Osher Center for Integrative Medicine, Department of Clinical Neuroscience, Karolinska Institutet, 17177 Stockholm, Sweden
• 2 Division for Psychology, Department of Clinical Neuroscience, Karolinska Institutet
• 3 Stress Research Institute, Stockholm University, Stockholm
• 4 Netherlands Institute for Neuroscience, an Institute of the Royal Netherlands Academy of Arts and Sciences, and VU Medical Center, Amsterdam, Netherlands
• Correspondence to: J Axelsson john.axelsson{at}ki.se
• Accepted 22 October 2010

Objective To investigate whether sleep deprived people are perceived as less healthy, less attractive, and more tired than after a normal night’s sleep.

Design Experimental study.

Setting Sleep laboratory in Stockholm, Sweden.

Participants 23 healthy, sleep deprived adults (age 18-31) who were photographed and 65 untrained observers (age 18-61) who rated the photographs.

Intervention Participants were photographed after a normal night’s sleep (eight hours) and after sleep deprivation (31 hours of wakefulness after a night of reduced sleep). The photographs were presented in a randomised order and rated by untrained observers.

Main outcome measure Difference in observer ratings of perceived health, attractiveness, and tiredness between sleep deprived and well rested participants using a visual analogue scale (100 mm).

Results Sleep deprived people were rated as less healthy (visual analogue scale scores, mean 63 (SE 2) v 68 (SE 2), P<0.001), more tired (53 (SE 3) v 44 (SE 3), P<0.001), and less attractive (38 (SE 2) v 40 (SE 2), P<0.001) than after a normal night’s sleep. The decrease in rated health was associated with ratings of increased tiredness and decreased attractiveness.

Conclusion Our findings show that sleep deprived people appear less healthy, less attractive, and more tired compared with when they are well rested. This suggests that humans are sensitive to sleep related facial cues, with potential implications for social and clinical judgments and behaviour. Studies are warranted for understanding how these effects may affect clinical decision making and can add knowledge with direct implications in a medical context.

Introduction

The recognition [of the case] depends in great measure on the accurate and rapid appreciation of small points in which the diseased differs from the healthy state Joseph Bell (1837-1911)

Good clinical judgment is an important skill in medical practice. This is well illustrated in the quote by Joseph Bell, 1 who demonstrated impressive observational and deductive skills. Bell was one of Sir Arthur Conan Doyle’s teachers and served as a model for the fictitious detective Sherlock Holmes. 2 Generally, human judgment involves complex processes, whereby ingrained, often less consciously deliberated responses from perceptual cues are mixed with semantic calculations to affect decision making. 3 Thus all social interactions, including diagnosis in clinical practice, are influenced by reflexive as well as reflective processes in human cognition and communication.

Sleep is an essential homeostatic process with well established effects on an individual’s physiological, cognitive, and behavioural functionality 4 5 6 7 and long term health, 8 but with only anecdotal support of a role in social perception, such as that underlying judgments of attractiveness and health. As illustrated by the common expression “beauty sleep,” an individual’s sleep history may play an integral part in the perception and judgments of his or her attractiveness and health. To date, the concept of beauty sleep has lacked scientific support, but the biological importance of sleep may have favoured a sensitivity to perceive sleep related cues in others. It seems warranted to explore such sensitivity, as sleep disorders and disturbed sleep are increasingly common in today’s 24 hour society and often coexist with some of the most common health problems, such as hypertension 9 10 and inflammatory conditions. 11

To describe the relation between sleep deprivation and perceived health and attractiveness we asked untrained observers to rate the faces of people who had been photographed after a normal night’s sleep and after a night of sleep deprivation. We chose facial photographs as the human face is the primary source of information in social communication. 12 A perceiver’s response to facial cues, signalling the bearer’s emotional state, intentions, and potential mate value, serves to guide actions in social contexts and may ultimately promote survival. 13 14 15 We hypothesised that untrained observers would perceive sleep deprived people as more tired, less healthy, and less attractive compared with after a normal night’s sleep.

Using an experimental design we photographed the faces of 23 adults (mean age 23, range 18-31 years, 11 women) between 14.00 and 15.00 under two conditions in a balanced design: after a normal night’s sleep (at least eight hours of sleep between 23.00-07.00 and seven hours of wakefulness) and after sleep deprivation (sleep 02.00-07.00 and 31 hours of wakefulness). We advertised for participants at four universities in the Stockholm area. Twenty of 44 potentially eligible people were excluded. Reasons for exclusion were reported sleep disturbances, abnormal sleep requirements (for example, sleep need out of the 7-9 hour range), health problems, or availability on study days (the main reason). We also excluded smokers and those who had consumed alcohol within two days of the protocol. One woman failed to participate in both conditions. Overall, we enrolled 12 women and 12 men.

The participants slept in their own homes. Sleep times were confirmed with sleep diaries and text messages. The sleep diaries (Karolinska sleep diary) included information on sleep latency, quality, duration, and sleepiness. Participants sent a text message to the research assistant by mobile phone (SMS) at bedtime and when they got up on the night before sleep deprivation. They had been instructed not to nap. During the normal sleep condition the participants’ mean duration of sleep, estimated from sleep diaries, was 8.45 (SE 0.20) hours. The sleep deprivation condition started with a restriction of sleep to five hours in bed; the participants sent text messages (SMS) when they went to sleep and when they woke up. The mean duration of sleep during this night, estimated from sleep diaries and text messages, was 5.06 (SE 0.04) hours. For the following night of total sleep deprivation, the participants were monitored in the sleep laboratory at all times. Thus, for the sleep deprivation condition, participants came to the laboratory at 22.00 (after 15 hours of wakefulness) to be monitored, and stayed awake for a further 16 hours. We therefore did not observe the participants during the first 15 hours of wakefulness, when they had had a slightly restricted sleep, but had good control over the last 16 hours of wakefulness when sleepiness increased in magnitude. For the sleep condition, participants came to the laboratory at 12.00 (after five hours of wakefulness). They were kept indoors two hours before being photographed to avoid the effects of exposure to sunlight and the weather. We had a series of five or six photographs (resolution 3872×2592 pixels) taken in a well lit room, with a constant white balance (×900l; colour temperature 4200 K, Nikon D80; Nikon, Tokyo). The white balance was differently set during the two days of the study and affected seven photographs (four taken during sleep deprivation and three during a normal night’s sleep). Removing these participants from the analyses did not affect the results. The distance from camera to head was fixed, as was the focal length, within 14 mm (between 44 and 58 mm). To ensure a fixed surface area of each face on the photograph, the focal length was adapted to the head size of each participant.

For the photo shoot, participants wore no makeup, had their hair loose (combed backwards if long), underwent similar cleaning or shaving procedures for both conditions, and were instructed to “sit with a straight back and look straight into the camera with a neutral, relaxed facial expression.” Although the photographer was not blinded to the sleep conditions, she followed a highly standardised procedure during each photo shoot, including minimal interaction with the participants. A blinded rater chose the most typical photograph from each series of photographs. This process resulted in 46 photographs; two (one from each sleep condition) of each of the 23 participants. This part of the study took place between June and September 2007.

In October 2007 the photographs were presented at a fixed interval of six seconds in a randomised order to 65 observers (mainly students at the Karolinska Institute, mean age 30 (range 18-61) years, 40 women), who were unaware of the conditions of the study. They rated the faces for attractiveness (very unattractive to very attractive), health (very sick to very healthy), and tiredness (not at all tired to very tired) on a 100 mm visual analogue scale. After every 23 photographs a brief intermission was allowed, including a working memory task lasting 23 seconds to prevent the faces being memorised. To ensure that the observers were not primed to tiredness when rating health and attractiveness they rated the photographs for attractiveness and health in the first two sessions and tiredness in the last. To avoid the influence of possible order effects we presented the photographs in a balanced order between conditions for each session.

Statistical analyses

Data were analysed using multilevel mixed effects linear regression, with two crossed independent random effects accounting for random variation between observers and participants using the xtmixed procedure in Stata 9.2. We present the effect of condition as a percentage of change from the baseline condition as the reference using the absolute value in millimetres (rated on the visual analogue scale). No data were missing in the analyses.

Sixty five observers rated each of the 46 photographs for attractiveness, health, and tiredness: 138 ratings by each observer and 2990 ratings for each of the three factors rated. When sleep deprived, people were rated as less healthy (visual analogue scale scores, mean 63 (SE 2) v 68 (SE 2)), more tired (53 (SE 3) v 44 (SE 3)), and less attractive (38 (SE 2) v 40 (SE 2); P<0.001 for all) than after a normal night’s sleep (table 1 ⇓ ). Compared with the normal sleep condition, perceptions of health and attractiveness in the sleep deprived condition decreased on average by 6% and 4% and tiredness increased by 19%.

 Multilevel mixed effects regression on effect of how sleep deprived people are perceived with respect to attractiveness, health, and tiredness

• View inline

A 10 mm increase in tiredness was associated with a −3.0 mm change in health, a 10 mm increase in health increased attractiveness by 2.4 mm, and a 10 mm increase in tiredness reduced attractiveness by 1.2 mm (table 2 ⇓ ). These findings were also presented as correlation, suggesting that faces with perceived attractiveness are positively associated with perceived health (r=0.42, fig 1 ⇓ ) and negatively with perceived tiredness (r=−0.28, fig 1). In addition, the average decrease (for each face) in attractiveness as a result of deprived sleep was associated with changes in tiredness (−0.53, n=23, P=0.03) and in health (0.50, n=23, P=0.01). Moreover, a strong negative association was found between the respective perceptions of tiredness and health (r=−0.54, fig 1). Figure 2 ⇓ shows an example of observer rated faces.

 Associations between health, tiredness, and attractiveness

Fig 1  Relations between health, tiredness, and attractiveness of 46 photographs (two each of 23 participants) rated by 65 observers on 100 mm visual analogue scales, with variation between observers removed using empirical Bayes’ estimates

• Download figure
• Open in new tab
• Download powerpoint

Fig 2  Participant after a normal night’s sleep (left) and after sleep deprivation (right). Faces were presented in a counterbalanced order

To evaluate the mediation effects of sleep loss on attractiveness and health, tiredness was added to the models presented in table 1 following recommendations. 16 The effect of sleep loss was significantly mediated by tiredness on both health (P<0.001) and attractiveness (P<0.001). When tiredness was added to the model (table 1) with an estimated coefficient of −2.9 (SE 0.1; P<0.001) the independent effect of sleep loss on health decreased from −4.2 to −1.8 (SE 0.5; P<0.001). The effect of sleep loss on attractiveness decreased from −1.6 (table 1) to −0.62 (SE 0.4; P=0.133), with tiredness estimated at −1.1 (SE 0.1; P<0.001). The same approach applied to the model of attractiveness and health (table 2), with a decrease in the association from 2.4 to 2.1 (SE 0.1; P<0.001) with tiredness estimated at −0.56 (SE 0.1; P<0.001).

Sleep deprived people are perceived as less attractive, less healthy, and more tired compared with when they are well rested. Apparent tiredness was strongly related to looking less healthy and less attractive, which was also supported by the mediating analyses, indicating that a large part of the found effects and relations on appearing healthy and attractive were mediated by looking tired. The fact that untrained observers detected the effects of sleep loss in others not only provides evidence for a perceptual ability not previously subjected to experimental control, but also supports the notion that sleep history gives rise to socially relevant signals that provide information about the bearer. The adaptiveness of an ability to detect sleep related facial cues resonates well with other research, showing that small deviations from the average sleep duration in the long term are associated with an increased risk of health problems and with a decreased longevity. 8 17 Indeed, even a few hours of sleep deprivation inflict an array of physiological changes, including neural, endocrinological, immunological, and cellular functioning, that if sustained are relevant for long term health. 7 18 19 20 Here, we show that such physiological changes are paralleled by detectable facial changes.

These results are related to photographs taken in an artificial setting and presented to the observers for only six seconds. It is likely that the effects reported here would be larger in real life person to person situations, when overt behaviour and interactions add further information. Blink interval and blink duration are known to be indicators of sleepiness, 21 and trained observers are able to evaluate reliably the drowsiness of drivers by watching their videotaped faces. 22 In addition, a few of the people were perceived as healthier, less tired, and more attractive during the sleep deprived condition. It remains to be evaluated in follow-up research whether this is due to random error noise in judgments, or associated with specific characteristics of observers or the sleep deprived people they judge. Nevertheless, we believe that the present findings can be generalised to a wide variety of settings, but further studies will have to investigate the impact on clinical studies and other social situations.

Importantly, our findings suggest a prominent role of sleep history in several domains of interpersonal perception and judgment, in which sleep history has previously not been considered of importance, such as in clinical judgment. In addition, because attractiveness motivates sexual behaviour, collaboration, and superior treatment, 13 sleep loss may have consequences in other social contexts. For example, it has been proposed that facial cues perceived as attractive are signals of good health and that this recognition has been selected evolutionarily to guide choice of mate and successful transmission of genes. 13 The fact that good sleep supports a healthy look and poor sleep the reverse may be of particular relevance in the medical setting, where health estimates are an essential part. It is possible that people with sleep disturbances, clinical or otherwise, would be judged as more unhealthy, whereas those who have had an unusually good night’s sleep may be perceived as rather healthy. Compared with the sleep deprivation used in the present investigation, further studies are needed to investigate the effects of less drastic acute reductions of sleep as well as long term clinical effects.

Conclusions

People are capable of detecting sleep loss related facial cues, and these cues modify judgments of another’s health and attractiveness. These conclusions agree well with existing models describing a link between sleep and good health, 18 23 as well as a link between attractiveness and health. 13 Future studies should focus on the relevance of these facial cues in clinical settings. These could investigate whether clinicians are better than the average population at detecting sleep or health related facial cues, and whether patients with a clinical diagnosis exhibit more tiredness and are less healthy looking than healthy people. Perhaps the more successful doctors are those who pick up on these details and act accordingly.

Taken together, our results provide important insights into judgments about health and attractiveness that are reminiscent of the anecdotal wisdom harboured in Bell’s words, and in the colloquial notion of “beauty sleep.”

What is already known on this topic

Short or disturbed sleep and fatigue constitute major risk factors for health and safety

Complaints of short or disturbed sleep are common among patients seeking healthcare

The human face is the main source of information for social signalling

What this study adds

The facial cues of sleep deprived people are sufficient for others to judge them as more tired, less healthy, and less attractive, lending the first scientific support to the concept of “beauty sleep”

By affecting doctors’ general perception of health, the sleep history of a patient may affect clinical decisions and diagnostic precision

Cite this as: BMJ 2010;341:c6614

We thank B Karshikoff for support with data acquisition and M Ingvar for comments on an earlier draft of the manuscript, both without compensation and working at the Department for Clinical Neuroscience, Karolinska Institutet, Sweden.

Contributors: JA designed the data collection, supervised and monitored data collection, wrote the statistical analysis plan, carried out the statistical analyses, obtained funding, drafted and revised the manuscript, and is guarantor. TS designed and carried out the data collection, cleaned the data, drafted, revised the manuscript, and had final approval of the manuscript. JA and TS contributed equally to the work. MI wrote the statistical analysis plan, carried out the statistical analyses, drafted the manuscript, and critically revised the manuscript. EJWVS provided statistical advice, advised on data handling, and critically revised the manuscript. AO provided advice on the methods and critically revised the manuscript. ML provided administrative support, drafted the manuscript, and critically revised the manuscript. All authors approved the final version of the manuscript.

Funding: This study was funded by the Swedish Society for Medical Research, Rut and Arvid Wolff’s Memory Fund, and the Osher Center for Integrative Medicine.

Competing interests: All authors have completed the Unified Competing Interest form at www.icmje.org/coi_disclosure.pdf (available on request from the corresponding author) and declare: no support from any company for the submitted work; no financial relationships with any companies that might have an interest in the submitted work in the previous 3 years; no other relationships or activities that could appear to have influenced the submitted work.

Ethical approval: This study was approved by the Karolinska Institutet’s ethical committee. Participants were compensated for their participation.

Participant consent: Participant’s consent obtained.

Data sharing: Statistical code and dataset of ratings are available from the corresponding author at john.axelsson{at}ki.se .

This is an open-access article distributed under the terms of the Creative Commons Attribution Non-commercial License, which permits use, distribution, and reproduction in any medium, provided the original work is properly cited, the use is non commercial and is otherwise in compliance with the license. See: http://creativecommons.org/licenses/by-nc/2.0/ and http://creativecommons.org/licenses/by-nc/2.0/legalcode .

• ↵ Deten A, Volz HC, Clamors S, Leiblein S, Briest W, Marx G, et al. Hematopoietic stem cells do not repair the infarcted mouse heart. Cardiovasc Res 2005 ; 65 : 52 -63. OpenUrl Abstract / FREE Full Text
• ↵ Doyle AC. The case-book of Sherlock Holmes: selected stories. Wordsworth, 1993.
• ↵ Lieberman MD, Gaunt R, Gilbert DT, Trope Y. Reflection and reflexion: a social cognitive neuroscience approach to attributional inference. Adv Exp Soc Psychol 2002 ; 34 : 199 -249. OpenUrl CrossRef
• ↵ Drummond SPA, Brown GG, Gillin JC, Stricker JL, Wong EC, Buxton RB. Altered brain response to verbal learning following sleep deprivation. Nature 2000 ; 403 : 655 -7. OpenUrl CrossRef PubMed
• ↵ Harrison Y, Horne JA. The impact of sleep deprivation on decision making: a review. J Exp Psychol Appl 2000 ; 6 : 236 -49. OpenUrl CrossRef PubMed Web of Science
• ↵ Huber R, Ghilardi MF, Massimini M, Tononi G. Local sleep and learning. Nature 2004 ; 430 : 78 -81. OpenUrl CrossRef PubMed Web of Science
• ↵ Spiegel K, Leproult R, Van Cauter E. Impact of sleep debt on metabolic and endocrine function. Lancet 1999 ; 354 : 1435 -9. OpenUrl CrossRef PubMed Web of Science
• ↵ Kripke DF, Garfinkel L, Wingard DL, Klauber MR, Marler MR. Mortality associated with sleep duration and insomnia. Arch Gen Psychiatry 2002 ; 59 : 131 -6. OpenUrl CrossRef PubMed Web of Science
• ↵ Olson LG, Ambrogetti A. Waking up to sleep disorders. Br J Hosp Med (Lond) 2006 ; 67 : 118 , 20. OpenUrl PubMed
• ↵ Rajaratnam SM, Arendt J. Health in a 24-h society. Lancet 2001 ; 358 : 999 -1005. OpenUrl CrossRef PubMed Web of Science
• ↵ Ranjbaran Z, Keefer L, Stepanski E, Farhadi A, Keshavarzian A. The relevance of sleep abnormalities to chronic inflammatory conditions. Inflamm Res 2007 ; 56 : 51 -7. OpenUrl CrossRef PubMed Web of Science
• ↵ Haxby JV, Hoffman EA, Gobbini MI. The distributed human neural system for face perception. Trends Cogn Sci 2000 ; 4 : 223 -33. OpenUrl CrossRef PubMed Web of Science
• ↵ Rhodes G. The evolutionary psychology of facial beauty. Annu Rev Psychol 2006 ; 57 : 199 -226. OpenUrl CrossRef PubMed Web of Science
• ↵ Todorov A, Mandisodza AN, Goren A, Hall CC. Inferences of competence from faces predict election outcomes. Science 2005 ; 308 : 1623 -6. OpenUrl Abstract / FREE Full Text
• ↵ Willis J, Todorov A. First impressions: making up your mind after a 100-ms exposure to a face. Psychol Sci 2006 ; 17 : 592 -8. OpenUrl Abstract / FREE Full Text
• ↵ Krull JL, MacKinnon DP. Multilevel modeling of individual and group level mediated effects. Multivariate Behav Res 2001 ; 36 : 249 -77. OpenUrl CrossRef Web of Science
• ↵ Ayas NT, White DP, Manson JE, Stampfer MJ, Speizer FE, Malhotra A, et al. A prospective study of sleep duration and coronary heart disease in women. Arch Intern Med 2003 ; 163 : 205 -9. OpenUrl CrossRef PubMed Web of Science
• ↵ Bryant PA, Trinder J, Curtis N. Sick and tired: does sleep have a vital role in the immune system. Nat Rev Immunol 2004 ; 4 : 457 -67. OpenUrl CrossRef PubMed Web of Science
• ↵ Cirelli C. Cellular consequences of sleep deprivation in the brain. Sleep Med Rev 2006 ; 10 : 307 -21. OpenUrl CrossRef PubMed Web of Science
• ↵ Irwin MR, Wang M, Campomayor CO, Collado-Hidalgo A, Cole S. Sleep deprivation and activation of morning levels of cellular and genomic markers of inflammation. Arch Intern Med 2006 ; 166 : 1756 -62. OpenUrl CrossRef PubMed Web of Science
• ↵ Schleicher R, Galley N, Briest S, Galley L. Blinks and saccades as indicators of fatigue in sleepiness warnings: looking tired? Ergonomics 2008 ; 51 : 982 -1010. OpenUrl CrossRef PubMed Web of Science
• ↵ Wierwille WW, Ellsworth LA. Evaluation of driver drowsiness by trained raters. Accid Anal Prev 1994 ; 26 : 571 -81. OpenUrl CrossRef PubMed Web of Science
• ↵ Horne J. Why we sleep—the functions of sleep in humans and other mammals. Oxford University Press, 1988.

Find anything you save across the site in your account

The Real Lesson of the Stanford Prison Experiment

On the morning of August 17, 1971, nine young men in the Palo Alto area received visits from local police officers. While their neighbors looked on, the men were arrested for violating Penal Codes 211 and 459 (armed robbery and burglary), searched, handcuffed, and led into the rear of a waiting police car. The cars took them to a Palo Alto police station, where the men were booked, fingerprinted, moved to a holding cell, and blindfolded. Finally, they were transported to the Stanford County Prison—also known as the Stanford University psychology department.

They were willing participants in the Stanford Prison Experiment, one of the most controversial studies in the history of social psychology. (It’s the subject of a new film of the same name—a drama, not a documentary—starring Billy Crudup, of “Almost Famous,” as the lead investigator, Philip Zimbardo. It opens July 17th.) The study subjects, middle-class college students, had answered a questionnaire about their family backgrounds, physical- and mental-health histories, and social behavior, and had been deemed “normal”; a coin flip divided them into prisoners and guards. According to the lore that’s grown up around the experiment, the guards, with little to no instruction, began humiliating and psychologically abusing the prisoners within twenty-four hours of the study’s start. The prisoners, in turn, became submissive and depersonalized, taking the abuse and saying little in protest. The behavior of all involved was so extreme that the experiment, which was meant to last two weeks, was terminated after six days.

Less than a decade earlier, the Milgram obedience study had shown that ordinary people, if encouraged by an authority figure, were willing to shock their fellow-citizens with what they believed to be painful and potentially lethal levels of electricity. To many, the Stanford experiment underscored those findings, revealing the ease with which regular people, if given too much power, could transform into ruthless oppressors. Today, more than forty-five years later, many look to the study to make sense of events like the behavior of the guards at Abu Ghraib and America’s epidemic of police brutality. The Stanford Prison Experiment is cited as evidence of the atavistic impulses that lurk within us all; it’s said to show that, with a little nudge, we could all become tyrants.

And yet the lessons of the Stanford Prison Experiment aren’t so clear-cut. From the beginning, the study has been haunted by ambiguity. Even as it suggests that ordinary people harbor ugly potentialities, it also testifies to the way our circumstances shape our behavior. Was the study about our individual fallibility, or about broken institutions? Were its findings about prisons, specifically, or about life in general? What did the Stanford Prison Experiment really show?

The appeal of the experiment has a lot to do with its apparently simple setup: prisoners, guards, a fake jail, and some ground rules. But, in reality, the Stanford County Prison was a heavily manipulated environment, and the guards and prisoners acted in ways that were largely predetermined by how their roles were presented. To understand the meaning of the experiment, you have to understand that it wasn’t a blank slate; from the start, its goal was to evoke the experience of working and living in a brutal jail.

From the first, the guards’ priorities were set by Zimbardo. In a presentation to his Stanford colleagues shortly after the study’s conclusion, he described the procedures surrounding each prisoner’s arrival: each man was stripped and searched, “deloused,” and then given a uniform—a numbered gown, which Zimbardo called a “dress,” with a heavy bolted chain near the ankle, loose-fitting rubber sandals, and a cap made from a woman’s nylon stocking. “Real male prisoners don't wear dresses,” Zimbardo explained, “but real male prisoners, we have learned, do feel humiliated, do feel emasculated, and we thought we could produce the same effects very quickly by putting men in a dress without any underclothes.” The stocking caps were in lieu of shaving the prisoner’s heads. (The guards wore khaki uniforms and were given whistles, nightsticks, and mirrored sunglasses inspired by a prison guard in the movie “Cool Hand Luke.”)

Often, the guards operated without explicit, moment-to-moment instructions. But that didn’t mean that they were fully autonomous: Zimbardo himself took part in the experiment, playing the role of the prison superintendent. (The prison’s “warden” was also a researcher.) /Occasionally, disputes between prisoner and guards got out of hand, violating an explicit injunction against physical force that both prisoners and guards had read prior to enrolling in the study. When the “superintendent” and “warden” overlooked these incidents, the message to the guards was clear: all is well; keep going as you are. The participants knew that an audience was watching, and so a lack of feedback could be read as tacit approval. And the sense of being watched may also have encouraged them to perform. Dave Eshelman, one of the guards, recalled that he “consciously created” his guard persona. “I was in all kinds of drama productions in high school and college. It was something I was very familiar with: to take on another personality before you step out on the stage,” Eshelman said. In fact, he continued, “I was kind of running my own experiment in there, by saying, ‘How far can I push these things and how much abuse will these people take before they say, ‘Knock it off?’ ”

Other, more subtle factors also shaped the experiment. It’s often said that the study participants were ordinary guys—and they were, indeed, determined to be “normal” and healthy by a battery of tests. But they were also a self-selected group who responded to a newspaper advertisement seeking volunteers for “a psychological study of prison life.” In a 2007 study, the psychologists Thomas Carnahan and Sam McFarland asked whether that wording itself may have stacked the odds. They recreated the original ad, and then ran a separate ad omitting the phrase “prison life.” They found that the people who responded to the two ads scored differently on a set of psychological tests. Those who thought that they would be participating in a prison study had significantly higher levels of aggressiveness, authoritarianism, Machiavellianism, narcissism, and social dominance, and they scored lower on measures of empathy and altruism.

Moreover, even within that self-selected sample, behavioral patterns were far from homogeneous. Much of the study’s cachet depends on the idea that the students responded en masse, giving up their individual identities to become submissive “prisoners” and tyrannical “guards.” But, in fact, the participants responded to the prison environment in all sorts of ways. While some guard shifts were especially cruel, others remained humane. Many of the supposedly passive prisoners rebelled. Richard Yacco, a prisoner, remembered “resisting what one guard was telling me to do and being willing to go into solitary confinement. As prisoners, we developed solidarity—we realized that we could join together and do passive resistance and cause some problems.”

What emerges from these details isn’t a perfectly lucid photograph but an ambiguous watercolor. While it’s true that some guards and prisoners behaved in alarming ways, it’s also the case that their environment was designed to encourage—and, in some cases, to require—those behaviors. Zimbardo himself has always been forthcoming about the details and the nature of his prison experiment: he thoroughly explained the setup in his original study and, in an early write-up , in which the experiment was described in broad strokes only, he pointed out that only “about a third of the guards became tyrannical in their arbitrary use of power.” (That’s about four people in total.) So how did the myth of the Stanford Prison Experiment—“Lord of the Flies” in the psych lab—come to diverge so profoundly from the reality?

In part, Zimbardo’s earliest statements about the experiment are to blame. In October, 1971, soon after the study’s completion—and before a single methodologically and analytically rigorous result had been published—Zimbardo was asked to testify before Congress about prison reform. His dramatic testimony , even as it clearly explained how the experiment worked, also allowed listeners to overlook how coercive the environment really was. He described the study as “an attempt to understand just what it means psychologically to be a prisoner or a prison guard.” But he also emphasized that the students in the study had been “the cream of the crop of this generation,” and said that the guards were given no specific instructions, and left free to make “up their own rules for maintaining law, order, and respect.” In explaining the results, he said that the “majority” of participants found themselves “no longer able to clearly differentiate between role-playing and self,” and that, in the six days the study took to unfold, “the experience of imprisonment undid, although temporarily, a lifetime of learning; human values were suspended, self-concepts were challenged, and the ugliest, most base, pathological side of human nature surfaced.” In describing another, related study and its implications for prison life, he said that “the mere act of assigning labels to people, calling some people prisoners and others guards, is sufficient to elicit pathological behavior.”

Zimbardo released video to NBC, which ran a feature on November 26, 1971. An article ran in the Times Magazine in April of 1973. In various ways, these accounts reiterated the claim that relatively small changes in circumstances could turn the best and brightest into monsters or depersonalized serfs. By the time Zimbardo published a formal paper about the study , in a 1973 issue of the International Journal of Crim__i__nology and Penology , a streamlined and unequivocal version of events had become entrenched in the national consciousness—so much so that a 1975 methodological critique fell largely on deaf ears.

Forty years later, Zimbardo still doesn’t shy away from popular attention. He served as a consultant on the new film, which follows his original study in detail, relying on direct transcripts from the experimental recordings and taking few dramatic liberties. In many ways, the film is critical of the study: Crudup plays Zimbardo as an overzealous researcher overstepping his bounds, trying to create a very specific outcome among the students he observes. The filmmakers even underscore the flimsiness of the experimental design, inserting characters who point out that Zimbardo is not a disinterested observer. They highlight a real-life conversation in which another psychologist asks Zimbardo whether he has an “independent variable.” In describing the study to his Stanford colleagues shortly after it ended, Zimbardo recalled that conversation: “To my surprise, I got really angry at him,” he said. “The security of my men and the stability of my prison was at stake, and I have to contend with this bleeding-heart, liberal, academic, effete dingdong whose only concern was for a ridiculous thing like an independent variable. The next thing he’d be asking me about was rehabilitation programs, the dummy! It wasn’t until sometime later that I realized how far into the experiment I was at that point.”

In a broad sense, the film reaffirms the opinion of John Mark, one of the guards, who, looking back, has said that Zimbardo’s interpretation of events was too shaped by his expectations to be meaningful: “He wanted to be able to say that college students, people from middle-class backgrounds ... will turn on each other just because they’re given a role and given power. Based on my experience, and what I saw and what I felt, I think that was a real stretch.”

If the Stanford Prison Experiment had simulated a less brutal environment, would the prisoners and guards have acted differently? In December, 2001 , two psychologists, Stephen Reicher and Alexander Haslam, tried to find out. They worked with the documentaries unit of the BBC to partially recreate Zimbardo’s setup over the course of an eight-day experiment. Their guards also had uniforms, and were given latitude to dole out rewards and punishments; their prisoners were placed in three-person cells that followed the layout of the Stanford County Jail almost exactly. The main difference was that, in this prison, the preset expectations were gone. The guards were asked to come up with rules prior to the prisoners’ arrival, and were told only to make the prison run smoothly. (The BBC Prison Study, as it came to be called, differed from the Stanford experiment in a few other ways, including prisoner dress; for a while, moreover, the prisoners were told that they could become guards through good behavior, although, on the third day, that offer was revoked, and the roles were made permanent.)

Within the first few days of the BBC study, it became clear that the guards weren’t cohering as a group. “Several guards were wary of assuming and exerting their authority,” the researchers wrote. The prisoners, on the other hand, developed a collective identity. In a change from the Stanford study, the psychologists asked each participant to complete a daily survey that measured the degree to which he felt solidarity with his group; it showed that, as the guards grew further apart, the prisoners were growing closer together. On the fourth day, three cellmates decided to test their luck. At lunchtime, one threw his plate down and demanded better food, another asked to smoke, and the third asked for medical attention for a blister on his foot. The guards became disorganized; one even offered the smoker a cigarette. Reicher and Haslam reported that, after the prisoners returned to their cells, they “literally danced with joy.” (“That was fucking sweet,” one prisoner remarked.) Soon, more prisoners began to challenge the guards. They acted out during roll call, complained about the food, and talked back. At the end of the sixth day, the three insubordinate cellmates broke out and occupied the guards’ quarters. “At this point,” the researchers wrote, “the guards’ regime was seen by all to be unworkable and at an end.”

Taken together, these two studies don’t suggest that we all have an innate capacity for tyranny or victimhood. Instead, they suggest that our behavior largely conforms to our preconceived expectations. All else being equal, we act as we think we’re expected to act—especially if that expectation comes from above. Suggest, as the Stanford setup did, that we should behave in stereotypical tough-guard fashion, and we strive to fit that role. Tell us, as the BBC experimenters did, that we shouldn’t give up hope of social mobility, and we act accordingly.

This understanding might seem to diminish the power of the Stanford Prison Experiment. But, in fact, it sharpens and clarifies the study’s meaning. Last weekend brought the tragic news of Kalief Browder’s suicide . At sixteen, Browder was arrested, in the Bronx, for allegedly stealing a backpack; after the arrest, he was imprisoned at Rikers for three years without trial . (Ultimately, the case against him was dismissed.) While at Rikers, Browder was the object of violence from both prisoners and guards, some of which was captured on video . It’s possible to think that prisons are the way they are because human nature tends toward the pathological. But the Stanford Prison Experiment suggests that extreme behavior flows from extreme institutions. Prisons aren’t blank slates. Guards do indeed self-select into their jobs, as Zimbardo’s students self-selected into a study of prison life. Like Zimbardo’s men, they are bombarded with expectations from the first and shaped by preëxisting norms and patterns of behavior. The lesson of Stanford isn’t that any random human being is capable of descending into sadism and tyranny. It’s that certain institutions and environments demand those behaviors—and, perhaps, can change them.

• Work & Careers
• Life & Arts

Inside Thatcher’s Monetarism Experiment — terminal velocity

To read this article for free, register now.

Once registered, you can: • Read free articles • Get our Editor's Digest and other newsletters • Follow topics and set up personalised events • Access Alphaville: our popular markets and finance blog

Explore more offers.

Then $75 per month. Complete digital access to quality FT journalism. Cancel anytime during your trial.

FT Digital Edition

Today's FT newspaper for easy reading on any device. This does not include ft.com or FT App access.

• Global news & analysis
• Expert opinion

Standard Digital

Essential digital access to quality FT journalism on any device. Pay a year upfront and save 20%.

• FT App on Android & iOS
• FT Edit app
• FirstFT: the day's biggest stories
• 20+ curated newsletters
• Follow topics & set alerts with myFT
• FT Videos & Podcasts

Terms & Conditions apply

Explore our full range of subscriptions.

Why the ft.

See why over a million readers pay to read the Financial Times.

• Share full article

Opinion Guest Essay

New York’s Promising Experiment in Making the Subways Safer

Credit... By Gus Aronson

Supported by

By Nicole Gelinas

Ms. Gelinas is a contributing editor for the Manhattan Institute’s City Journal. Her forthcoming book is “Movement: New York’s Long War to Take Back Its Streets From the Car.”

• Aug. 14, 2024

Ameed Ademolu, a close-cropped and bespectacled nurse, descends the street-level stairs to the subway, clad in an orange and yellow vest and carrying a clipboard, paper and pencil. He halts, reaches into a pocket to unfold a KN95 mask and fits it over his mouth and nose. “The mask is for me,” he says, because “I’ve had people spit in my face.”

At the Fulton Street station in downtown Manhattan, Mr. Ademolu encounters two men, one in his 20s and one older, the latter dependent on a cart piled with his belongings as a walker. The younger man is coherent but agitated, ripping up newspaper and scattering the pieces as he puffs on his vape. Both are homeless and have become friends after months on the subways.

Mr. Ademolu talks to them for half an hour. The older man declines Mr. Ademolu’s offer to have a shelter van pick him up, saying he knows that the shelter system cannot easily accommodate someone who cannot walk up and down stairs. The younger man decides he will go to a shelter. Mr. Ademolu takes the man upstairs to await the van.

Mr. Ademolu spends his days like this, pacing Manhattan subway platforms and stations, searching for people who appear severely mentally ill. He is a member of a Subway Co-Response Outreach, or SCOUT, team, a new project between the city and the state-run Metropolitan Transportation Authority.

His work represents the newer of two strategies that New York’s city and state governments are using to dispel the atmosphere of disorder and danger that permeates the subways. The other is more old-fashioned policing and prosecution. Success for both depends on New York creating a functional mental health system and recreating a functional criminal justice system. If this approach works, it could become a model for other American cities struggling with crime and mental illness.

We are having trouble retrieving the article content.

Please enable JavaScript in your browser settings.

Thank you for your patience while we verify access. If you are in Reader mode please exit and  log into  your Times account, or  subscribe  for all of The Times.

Thank you for your patience while we verify access.

Already a subscriber?  Log in .

Want all of The Times?  Subscribe .

Advertisement

6 leadership lessons from Brian Niccol, the new Starbucks CEO who revived Chipotle

• Chipotle boss Brian Niccol will take over as CEO of Starbucks next month.
• He laid out his leadership style and approach to turning around companies on a podcast this year.
• Niccol follows trends, welcomes new ideas and feedback, prizes focus — and demands excellence.

Brian Niccol's decision to leave Chipotle and become CEO of Starbucks has sparked speculation about the changes he might make at the embattled coffee chain.

Niccol laid out his management approach on the "How Leaders Lead with David Novak" podcast earlier this year. He shared how he revitalizes struggling businesses, finds and introduces fresh ideas, and delivers world-beating products and services.

The burrito boss said he wants to be seen as a decisive CEO who nurtures future leaders and can grow a company, not just turn one around. He also broke down six key elements of his leadership style that he's likely to employ at Starbucks when he starts in September.

1. Tapped in

When you're trying to keep a brand relevant, it helps to be attuned to what's popular among younger generations.

Niccol is an active user of TikTok, Snapchat, and Instagram for that reason. He also ensures that Chipotle's bosses discuss the trends they're seeing and what their kids and their kids' friends are talking about every week.

"We want to be very much in touch with what young people are doing," Niccol said. "It gives us some insight into what are the things we can do to keep Chipotle fresh, keep Chipotle connected."

2. Embrace experiments

Leaders who welcome fresh thinking and experimentation have an edge in attracting talent and driving innovation.

"We are a place for ideas," Niccol said. "If you want to come to a place and you're willing to pressure test your ideas, we're going to give you a shot with those ideas to make it to market."

He ticked off several ideas that came from within Chipotle, including its metaverse experiments, the use of its reward program to bolster engagement with customers, and the addition of carne asada to menus.

Niccol said he uses a "stage gate process" to vet ideas and take them from concept to nationwide rollout, which includes performance metrics and store feedback.

Related stories

3. Listen and learn

Niccol shared the first step of his turnaround plans: "Take the time to understand who these people are."

Before he starts barking instructions or declaring what the company needs to do, he finds out about his new colleagues, their beliefs, and what's working or not working in their view.

He also emphasized the importance of reacting to customer and worker feedback. Soon after joining Chipotle in 2018, he visited some of the chain's restaurants and asked the staff what common complaints they hear.

A big one was the absence of a large portion of guacamole from the menu, which forced workers to invent a workaround.

"They were just using, I think it was the queso cup, and they would charge people two smalls," Niccol said. "And I'm like, that's ridiculous."

Niccol promptly resolved the problem by adding the necessary buttons to cashier terminals and containers to serving lines.

4. Build the right team

Once Niccol gets to know his new coworkers and identifies the pressing problems, he decides who he wants on his team and who isn't a good fit.

"I believe in you. Hopefully, you believe in me. Let's lock arms and do this," he said, giving a sense of how he gets the people he chooses to commit to his strategy.

Niccol also figures out the personnel gaps that need filling. At Chipotle, he realized he needed to find "great" marketing and supply chain leaders and expand the digital team to bolster its capabilities.

On the other hand, he moved the company's restaurant support center from Denver to Los Angeles because it needed a "complete reset."

5. Focus and action

Niccol likes to choose one or two things for his new team to tackle that will make an immediate difference.

At Chipotle, his initial priorities were catering to online orders with digital make lines and mobile pickup shelves, and improving how the restaurants operated and food was prepared and served.

He also likes to pick one or two things for each of his top managers to focus on. For example, he told Chipotle's marketing boss that he wanted the brand to be relevant and to get the word out about what makes customers love it.

"Collectively sharing the obligation to achieve those things is what I've done, and it appears to be effective," he said.

6. Demand the best

Niccol recalled that after Chipotle reopened following the pandemic, it was falling short of its previous high standards.

"I'd go into restaurants, we'd be out of chips. I would go into a restaurant near closing, we'd be out of chicken, rice," he said. Throughput, or number of transactions every 15 minutes, was "way off" pre-pandemic levels.

"It was just like, we had made way too many excuses for too long, and COVID gave us the ability to provide those excuses," he said. "And it's like, hey, it's time to put those aside. Let's get back to the standard of excellence that we know we can hit."

"I went to our team director meeting and I said, 'Hey gang, I am here to tell you we are not succeeding,'" Niccol said. "We collectively should be embarrassed that we have a report telling us when we're out of chips."

"Like, we're a Mexican restaurant. When was the last time you went to a restaurant and they didn't have chips, right?" Niccol added, noting Chipotle solved those issues by the next year after his intervention.

"Every once in a while, they respect you for recognizing we're not hitting the number."

Watch: Building communities inside and outside the company is critical to success for marketing leaders says Diana Frost, chief growth officer at Kraft Heinz

• Main content