Author: Cathleen O'Grady

Thoughts don't just flit around in our heads unobserved: humans know when something's going on in our own brains, and we evaluate our own thoughts. For example, we can judge when we're not certain about something, and act accordingly. This ability, called metacognition (thinking about thinking), has been found in a number of species, but humans are unusual in our ability to communicate what we know about our own thoughts and knowledge.

How early in life do we develop metacognition? Toddlers, who confidently proclaim knowledge of things they can’t possibly know, seem to be pretty bad at it. Babies, on the other hand, point at things to ask questions about them. They shouldn't be able to do this unless they've worked out that they don't know something.

It’s possible that previous experiments haven’t found evidence of metacognition in younger children because they just weren’t testing it in the right way. After all, other species have metacognition, and experimenters have found ways to test that even though the animals can’t talk about what they know. What if children under four years old experience and use metacognition, but are just bad at realizing it and letting anyone know?

Approaches from a number of behavioral sciences can be used to understand human honesty. They’re not bothered with making particular judgements about whether a behavior is good or bad, but rather these approaches focus on understanding why people behave honestly or dishonestly in different situations.

Two economists at the University of Nottingham, Simon Gächter and Jonathan Schulz, have published an intriguing suggestion about the roots of dishonesty: they suggest that a corrupt social environment, rife with political corruption and tax evasion, can trickle down to the individual level and make people in such a country more likely to be dishonest in some contexts. Based on data gathering and behavioral experiments done in 23 countries, they found that people in more corrupt countries were more likely to cheat during an experiment.

The question was why—do national tendencies push the population toward more or less honesty, or do individuals drive the national tendency?

Studying people’s trust in robots is an academic field, but it's one that’s growing in relevance as we embrace a future of driverless cars and ever-more-powerful artificial intelligence. If we based our expectations on what we see in science fiction, we might expect that people have a profound mistrust in robots. Instead, research from Georgia Institute of Technology has found that it’s possible that we could face the problem of trusting robots way too much.

The researchers conducted a study that they will be presenting next week at an international conference on robot-human interaction, so the full paper hasn’t yet been published. However, an early press release and preliminary paper give some of the details of the study, which initially set out to find out whether high-rise occupants would be likely to trust a robot’s instructions in an evacuation scenario. The researchers were concerned with what robot behavior would win or lose people's trust.

The 26 participants used in the experiment had no idea what it was about; they were just asked to follow a robot that had the words “Emergency Guide Robot” printed prominently on its side. The first thing the robot was supposed to do was lead them to a room where they would read an article and take part in a survey (all as a distraction from the real task).

Camera trap footage, taken with no humans present to cause a disturbance, shows one chimp after another pick up a rock and hurl it at the same tree. Rocks pile up at the foot of the tree, which starts to show signs of wear and tear. For some reason, the chimps have picked this particular tree for an accumulation of hurled rocks.

“It was unlike anything I had ever observed among wild chimpanzees,” said primatologist Ammie Kalan. Her team has discovered the behavior in four distinct populations, and it’s possible that more will turn up as they continue searching. What the rock piles mean is an open question, but the discovery of such a distinct and puzzling stone tool use is unquestionably exciting.

We’ve known for a while now that chimps use stone tools for more easily discernible purposes, like extracting food. They’ve also been observed throwing objects, like rocks and branches, haphazardly. What’s different about this behavior is the consistent location and the repetitive actions seen in multiple individuals, which is what has led the researchers to label it as tool use. “It’s definitely the first indication of something else going on other than stone tool use for food extraction or for throwing rocks… haphazardly,” Kalan told Ars.

People often make assumptions about each other’s character traits based on facial features. While, to a limited extent, it seems like we can actually glean some reliable info about personality traits from someone’s face, we can also add a heaping pile of bias on top of that too. A paper in PLOS One explores one of these biases: the more attractive we think a person is, the more we’re inclined to think they’re intelligent. This “attractiveness halo” means that we’re likely to overestimate the intelligence of people we find attractive.

One of the complications in assessing something like this is that people vary in what they believe “intelligence” is. This is an entirely separate question from what any scientific consensus actually says about the concept of intelligence and how we measure it. Regardless of what science says, people will still have their own understanding of what the idea means, believing that intelligence is the result of things like a “growth mindset,” of conscientiousness, or of genetics.

This makes it a bit difficult to do an experiment. These different definitions of intelligence, the authors write, suggest that people will consider different things in a face to signal intelligence, leading to “less accurate perceptions of intelligence.” So just asking for people to rate the intelligence of a series of faces would be unlikely to identify our perceptions with a huge amount of accuracy. To account for this, the researchers decided to also ask about other ratings that could capture various beliefs about intelligence: conscientiousness, and academic performance.

There’s a deep sense of irony in adding to a never-ending series of headlines on a study that shouldn’t have had any attention paid to it at all. But the publication on the dangers of the “paleo” diet that’s spawned countless headlines is so flawed that it’s worth exploring why it got so much attention.

“Diabetes expert warns paleo diet is dangerous and increases weight gain,” claims the press release issued to promote a paper in last week’s Nature Nutrition and Diabetes. What's particularly flabbergasting about this situation is that the study didn’t actually have anything to do with a paleo diet—it used a low-carb, high-fat (LCHF) diet. The two diets can be similar—paleo diets tend to be lower in carbs and higher in fats—but they’re not the same thing, and it’s possible to eat a paleo diet that’s not high in fat.

The headline also doesn’t mention that the study looked at mice. But not many mice—only seventeen male New Zealand Obese (NZO) mice, which are inbred to be severely predisposed to obesity and diabetes. Given that the study wanted to find out whether the diet is suitable for obese, pre-diabetic people, it makes sense to study mice with a predisposition toward these conditions. But it does mean that the results don't apply more broadly. “It’s not even applicable to all mice, let alone all humans,” says Yoni Freedhoff, an assistant professor at the University of Ottawa with a specialty in debunking diet nonsense.

Computers have started to get really good at visual recognition. They can sometimes rival humans at recognizing the objects in a series of images. But does the similar end result mean that computers are mimicking the human visual system? Answering that question would indicate if there are still some areas where computer systems can't keep up with humans.

So, a new PNAS paper takes a look at just how different computer and human visual systems are.

The difference really boils down to the flexibility that human brains have and computers don’t. It’s much the same problem that speech recognition system face: humans can figure out that a mangled word “meant” something recognizable while a computer can’t. Likewise with images: humans can piece together what a blurry image might depict based on small clues in the picture, where a computer would be at a loss.

The idea behind standardized testing is that everyone gets the chance to perform on the same test in the same circumstances. In an ideal world, this should create a system where everyone’s test results are a good indicator of their skills, learning, and hard work.

The reality, of course, is different. Standardized testing faces a host of criticisms, some more valid than others. But even if we assume that everyone walks into a standardized test with the same background, when the test happens matters. The timing of the test itself can have a marked impact on student scores, according to a new paper in PNAS.

The paper found that the later in the day a standardized test was held, the lower the scores were. That’s an important finding, given how much rests on standardized test results. These tests not only form the basis of education policy in countries all over the world, but they're often also used to decide how funding should be distributed among schools. And, most obvious of all, a test score can determine the course of a student’s life.

The level of altruism that humans display is an anomaly in the animal world. Most species don’t interact peacefully with strangers every day or build large, stable societies that rely on cooperative behavior between unrelated individuals. Although there are animals that show altruistic behavior toward their relatives or breeding partners, we still don’t know how humans managed to develop the extreme level of cooperation between strangers needed to build and maintain our societies.

A paper in Nature hints that religion may be one of the keys to understanding this cooperation. The paper's authors suggest that, as people started to believe in gods who see everything and punish wrongdoing, they may have had more motivation to behave nicely toward strangers. They also suggest that beliefs in more powerful gods might widen the circle of cooperation: the more all-knowing your deity, the farther away people can be from you and still benefit from your cooperation.

To test this idea, the authors studied nearly 600 people with a wide range of beliefs from countries around the world. The beliefs included predominant world religions such as Christianity and Hinduism but also local traditions like ancestor worship, animism, and belief in supernatural entities like saints or ghosts. After answering detailed questions about what they believed, participants played a game to assess how they would act toward other people.

A paper in Monday's issue of PNAS reports that sleep-deprived people are up to 4.5 times more likely to sign a false confession. There’s an important weakness in the experiment, however, in that participants didn’t face any penalty for signing the confession. But the study does tie in with other evidence suggesting that specific interrogation practices can lead to false confessions, so it may be an important chunk of pixels in an emerging picture.

As it stands, there's evidence that sleep deprivation interferes with people's ability to make rational decisions. There's also evidence that most false confessions are signed after interrogations that lasted more than 12 hours. Taken together, these findings suggest that sleep deprivation could play a role in how an interrogation turns out but doesn't tell us anything about whether this does happen.

The PNAS paper suggests that sleepier people may be more likely to falsely confess but that people's individual characteristics also play a role: people who show a more impulsive decision-making approach are more likely to sign.