Zachary del Rosario

Zachary del Rosario

Probability underpins AI, cryptography and statistics. However, as the philosopher Bertrand Russell said, “Probability is the most important concept in modern science, especially as nobody has the slightest notion what it means.”

I teach statistics to engineers, so I know that while probability is important, it is counterintuitive.

Probability is a branch of mathematics that describes randomness. When scientists describe randomness, they’re describing chance events – like a coin flip – not strange occurrences, like a person dressed as a zebra. While scientists do not have a way to predict strange occurrences, probability does predict long-run behavior – that is, the trends that emerge from many repeated events.

Mathematics, Education, Explainer, Statistics, Probability, Frequency, Doing science
Photo credit: Zebras in La Paz, Bolivia by EEJCC, Own Work CC A-SA 4.0; CC BY-SA – We may say ‘random’ to describe strange occurrences (person dressed as zebra), but probability describes chance events (a coin flip).

Modeling with probability

Since probability is about events, a scientist must choose which events to study. This choice defines the sample space. When flipping a coin, for example, you might define your event as the way it lands.

Coins almost always land on heads or tails. However, it’s possible – if very unlikely – for a coin to land on its side. So to create a sample space, you’d have two choices: heads and tails, or heads, tails and side. For now, ignore the side landings and use heads and tails as our sample space.

Next, you would assign probabilities to the events. Probability describes the rate of occurrence of an event and takes values between 0% and 100%. For example, a fair flip will tend to land 50% heads up and 50% tails up.

To assign probabilities, however, you need to think carefully about the scenario. What if the person flipping the coin is a cheater? There’s a sneaky technique to “wobble” the coin without flipping, controlling the outcome. Even if you can prevent cheating, real coin flips are slightly more probable to land on their starting face – so if you start the flip with the coin heads up, it’s very slightly more likely to land heads up.

In both the cheating and real flip cases, you need an appropriate sample space: starting face and other face. To have a fair flip in the real world, you’d need an additional step where you randomly – with equal probability – choose the starting face, then flip the coin.

Mathematics, Education, Explainer, Statistics, Probability, Frequency, Doing science
Photo credit: Zachary del Rosario, CC BY-SA – The probabilities for different coin-flipping scenarios.

These assumptions add up quickly. To have a fair flip, you had to ignore side landings, assume no one is cheating, and assume the starting face is evenly random. Together, these assumptions constitute a model for the coin flip with random outcomes. Probability tells us about the long-run behavior of a random model. In the case of the coin model, probability describes how many coins land on heads out of many flips.

But instead of using a random model, why not just solve the coin toss using physics? Actually, scientists have done just that, and the physics shows that slight changes in the speed of the flip determine whether it comes up heads or tails. This sensitivity makes a coin flip unpredictable, so a random model is a good one.

Frequency vs. probability

Probability differs from frequency, which is the rate of events in a sequence. For example, if you flip a coin eight times and get two heads, that’s a frequency of 25%. Even if the probability of flipping a coin and seeing heads is 50% over the long run, each short sequence of flips will come out different. Four heads and four tails is the most probable outcome from eight flips, but other events can – and will – happen.

Frequency and probability are the same in one special setting: when the number of data points goes to infinity. In this sense, probability tells us about long-run behavior.

Mathematics, Education, Explainer, Statistics, Probability, Frequency, Doing science
Photo credit: Zachary del Rosario, CC BY-SA – Probabilities for all possible outcomes of eight ‘fair’ coin flips.

Applications to AI, cryptography and statistics

Probability isn’t just useful for predicting coin flips. It underlies many modern technological systems.

For example, AI systems such as large language models, or LLMs, are based on next-word prediction. Essentially, they compute a probability for the words that follow your prompt. For example, with the prompt “New York” you might get “City” or “State” as the predicted next word, because in the training data those are the words that most frequently follow.

But since probability describes randomness, the outputs of a LLM are random. Just like a sequence of coin flips is not guaranteed to come out the same way every time, if you ask an LLM the same question again, you will tend to get a different response. Effectively, each next word is treated like a new coin flip.

Randomness is also key to cryptography: the science of securing information. Cryptographic communication uses a shared secret, such as a password, to secure information. However, surprising randomness isn’t good enough for security, which is why picking a surprising word is a bad choice of password. A shared secret is only secure if it’s hard to guess. Even if a word is surprising, real words are easier to guess than flipping a “coin” for each letter.

You can make a much stronger password by using probability to choose characters at random on your keyboard – or better yet, use a password manager.

Finally, randomness is key in statistics. Statisticians are responsible for designing and analyzing studies to make use of limited data. This practice is especially important when studying medical treatments, because every data point represents a person’s life.

The gold standard is a randomized controlled trial. Participants are assigned to receive the new treatment or the current standard of care based on a fair coin flip. It may seem strange to do this assignment randomly – using coin flips to make decisions about lives. However, the unpredictability serves an important role, as it ensures that nothing about the person affects their chance to get the treatment: not age, gender, race, income or any other factor. The unpredictability helps scientists ensure that only the treatment causes the observed result and not any other factor.

So what does probability mean? Like any kind of math, it’s only a model, meaning it can’t perfectly describe the world. In the examples discussed, probability is useful for describing long-term behaviors and using unpredictability to solve practical problems.

This article originally appeared on The Conversation. You can read it here.

  • Too many males to choose from can overwhelm female frogs – and change who gets to mate
    There can be too many frogs in the dating pool. R. Andrew Odum/Photodisc via Getty Images

    Assistant Professor of Ecology & Evolutionary Biology, University of Tennessee; Smithsonian Institution

    If you’ve ever felt overwhelmed by menu options at a restaurant, cereal varieties at the grocery store or potential partners while scrolling through dating apps, you’ve experienced choice overload.

    People aren’t the only ones to have trouble making decisions when confronted with many options simultaneously. In our new study published in the journal iScience, my colleagues and I found similar effects of choice overload in female gray tree frogs when faced with too many calls from males looking for love.

    Scientists have observed many species having similar decision-making problems when experiencing choice overload. But we know little about how these kinds of cognitive constraints affect animals in natural settings.

    Indecision on the frog dating scene

    The gray tree frogs we focused on live in wooded areas near sources of water in North America. During the mating season, males gather in choruses and call to attract mates. Females pick mates based on several call attributes, including call length. Longer calls are more attractive.

    For our experiment, we captured 60 female tree frogs from breeding choruses in East Tennessee and brought them into our lab. Then we played recorded mating calls for each female, giving her a choice between one long, attractive “target” call and one, three or seven shorter, less attractive “distractor” calls.

    In the wild, a tree frog breeding chorus can include dozens or even hundreds of calling males. In our lab setup, the calls played sequentially, so they never overlapped.

    Females would hop toward the speaker broadcasting the call she preferred. As the number of distractor calls increased, females got worse at picking the attractive call, took longer to decide, and were more likely to defer choosing within the 10-minute window altogether.

    Black and white aerial view of a circular area with a white star in the middle
    View from above of the frog test arena, where researchers would place a female frog in the center and then play various male mating calls from speakers placed around the perimeter. Jessie Tanner

    Cognitive confusion

    Because we eliminated call overlap entirely, we know this decline wasn’t due to what researchers call energetic masking. That’s the well-documented problem of overlapping signals impairing an individual’s ability to detect and discriminate a signal from noise.

    Biologists have long known that female preferences can weaken in choruses and have usually assumed the explanation is energetic masking: Too many overlapping signals create a challenge for the frogs’ perceptual systems. In other words, they can’t detect or discriminate individual calls.

    Instead, our study shows that a female’s own cognitive limits can make it hard for her to select her preferred mate. The challenge comes from too many males to choose from, not being unable to hear the males clearly.

    When females can’t make the optimal choice

    Call traits are linked to offspring fitness in gray tree frogs. If females have trouble recognizing the calls they prefer in a real chorus, they aren’t able to pick the mates with the best genes to pass on. They may end up mating with males they would otherwise reject. These males with less going for them then pass on their less desirable traits, which could have negative consequences for the offspring.

    Smaller frog on top of larger frog with grass blades around
    Females’ cognitive overload could be an opportunity for males who might not otherwise get a chance to mate. Jessie Tanner

    Our finding raises the possibility that males could exploit this cognitive limitation in females. Less attractive males might do better by joining bigger choruses, where the sheer number of competitors makes it harder for females to discriminate against them. Conversely, more attractive males may benefit from calling in smaller groups, or even in isolation. My colleagues and I hope to investigate some of these possibilities soon.

    More options, more opportunities … to a point

    We ran our experiments in a simplified lab arena, which allowed us to isolate choice overload from other effects. We removed the real-world clutter of overlapping calls, vegetation and risk from predators that females would face in an actual pond. We also limited female choice to eight calls, though frogs may hear even more in natural settings.

    The next step in this research is figuring out how much choice overload matters in a real chorus, where calls do overlap, and where sensory processing and cognitive processing are probably both at play. We’re particularly interested in seeing whether, in natural choruses, decisions do take longer and females really end up mating with less preferable males.

    Choice overload isn’t unique to frogs, or even to mate choice. Researchers have seen bees choosing among flowers, ants comparing potential nest sites and mosquitoes selecting egg-laying sites all making worse decisions as the number of competing options increases. And recent research on field crickets and tree frogs has found similar patterns in these animals as they pick mates based on their calls.

    Together, this growing body of evidence suggests that cognitive constraints are a general feature of animal decision-making.

    This article originally appeared on The Conversation. You can read it here.

  • Why do frozen turkeys explode when deep-fried?
    When water and boiling oil mix, the result can be explosive, as seen in this demonstration. U.S. Navy photo by Mass Communication Specialist 2nd Class Travis Alston/Released via Flickr
    ,

    Why do frozen turkeys explode when deep-fried?

    A holiday kitchen disaster comes down to some powerful chemistry.

    Associate Professor of Chemistry, University of Richmond

    Deep-frying a turkey is a great way to get a delicious, moist meal for Thanksgiving. But this method of cooking can be a very dangerous undertaking.

    Every fall, millions of dollars of damage, trips to the ER and even deaths result from attempts to deep-fry turkeys. The vast majority of these accidents happen because people put frozen turkeys into boiling oil. If you are considering deep-frying this year, do not forget to thaw and dry your turkey before placing it in the pot. Failure to do so may lead to an explosive disaster.

    What is so dangerous about putting even a partially frozen turkey in a deep-fryer?

    I’m a chemist who studies plant, fungal and animal compounds and have a love of food chemistry. The reason frozen turkeys explode, at its core, has to do with differences in density. There is a difference in density between oil and water and differences in the density of water between its solid, liquid and gas states. When these density differences interact in just the right way, you get an explosion.

    A cup filled with a layer of oil sitting on top of a layer of water.
    If water and oil are in the same container, the denser water will sink to the bottom. Daniel Merino, CC BY-ND

    Understanding density

    Density is how much an object weighs given a specific volume. For example, imagine you held an ice cube in one hand and a marshmallow in the other. While they are roughly the same size, the ice cube is heavier: It is more dense.

    The first important density difference when it comes to frying is that water is more dense than oil. This has to do with how tightly the molecules of each substance pack together and how heavy the atoms are that make up each liquid.

    Water molecules are small and pack tightly together. Oil molecules are much larger and don’t pack together as well by comparison. Additionally, water is composed of oxygen and hydrogen atoms, while oils are predominantly carbon and hydrogen. Oxygen is heavier than carbon. This means that, for example, one cup of water has more atoms than one cup of oil, and those individuals atoms are heavier. This is why oil floats on top of water. It is less dense.

    While different materials have different densities, liquids, solids and gases of a single material can have different densities as well. You observe this every time you place an ice cube in a glass of water: The ice floats to the top because it is less dense than water.

    When water absorbs heat, it changes to its gas phase, steam. Steam occupies 1,700 times the volume as the same number of liquid water molecules. You observe this effect when you boil water in a tea kettle. The force of expanding gas pushes steam out of the kettle through the whistle, causing the squealing noise.

    Frozen turkeys are filled with water

    A man in a fire suit placing a turkey in a fryer.
    Deep-frying a turkey can be safe as long as the turkey is thawed and dried. YinYang/E+ via Getty Images

    Frozen turkeys – or any kind of frozen meats, for that matter – contain a lot of ice. Raw meat can be anywhere from 56% to 73% water. If you have ever thawed a frozen piece of meat, you have probably seen all the liquid that comes out.

    For deep-frying, cooking oil is heated to around 350 degrees Fahrenheit (175 C). This is much hotter than the boiling point of water, which is 212 F (100 C). So when the ice in a frozen turkey comes in contact with the hot oil, the surface ice quickly turns to steam.

    This quick transition is not a problem when it happens at the very surface of the oil. The steam escapes harmlessly into the air.

    However, when you submerge a turkey into the oil, the ice inside the turkey absorbs the heat and melts, forming liquid water. Here is where the density comes into play.

    This liquid water is more dense than the oil, so it falls the bottom of the pot. The water molecules continue to absorb heat and energy and eventually they change phases and become steam. The water molecules then rapidly spread far apart from one another and the volume expands by 1,700 times. This expansion causes the density of the water to drop to a fraction of a percent of the density of the oil, so the gas wants to quickly rise to the surface.

    Combine the fast change in density together with the expansion of volume and you get an explosion. The steam expands and rises, blowing the boiling oil out the pot. If that weren’t dangerous enough, as the displaced oil comes into contact with a burner or flame, it can catch fire. Once some droplets of oil catch on fire, the flames will quickly ignite nearby oil molecules, resulting in a fast-moving and often catastrophic fire.

    Every year, thousands of accidents like this happen. So, should you decide to deep-fry a turkey for this year’s Thanksgiving, be sure to thoroughly thaw it and pat it dry. And next time you add a bit of liquid to an oil-filled pan and end up with oil all over the stove, you’ll know the science of why.

    This article originally appeared on The Conversation. You can read it here.

  • Tiny backpacks reveal the secret, sometimes circuitous migrations of small songbirds
    A tracker must be light enough to not weigh down a bird that weighs only as much as a spoonful of sugar. Thor Veen
    ,

    Tiny backpacks reveal the secret, sometimes circuitous migrations of small songbirds

    Lightweight sensors are revealing where small birds travel, and why their routes can be far longer than researchers expected.

    James Smithson Postdoctoral Fellow, Smithsonian Institution

    Squinting through my binoculars, I scanned the dense branches of a black spruce, trying my best to catch a glimpse of the thin legs of a small black, white and yellow songbird. I was on the hunt for a yellow-rumped warbler – a bird affectionately called “butter-butt” by bird-watchers after the distinctive buttery yellow patch of feathers above its tail.

    A woman in a veiled hat holds a small bird with red and yellow bands on its legs.
    Stephanie Szarmach holds a yellow-rumped warbler that was banded as part of a study. Johanna Beam

    Once I spotted a flash of red on one of the bird’s legs, I knew that this individual was special – in fact, it was one that I already knew well. That’s because I had placed identifying bands on it the previous year as part of my Ph.D. research, and it was carrying a very important piece of new technology.

    I played the high pitched “warble” of its song through a speaker placed below a mist net of fine, nearly invisible mesh. Thinking another male had intruded on his territory, the bird quickly descended. My tactic worked – he got caught in the net. As I gently removed him, I breathed a sigh of relief as I saw the Tic Tac-size device sitting on his back.

    A hand holding a bird that has a small grey device on its back
    A yellow-rumped warbler wears a multisensor geolocator backpack. Stephanie Szarmach

    Over the past year, this warbler carried a multisensor geolocator backpack throughout his migratory journey, recording the long trek from his breeding grounds in Anchorage, Alaska, to a yet unknown winter refuge and back again. Working with my Ph.D. adviser, evolutionary biologist David Toews, I deployed dozens of these cutting-edge tags to illuminate the migratory journeys of these high-latitude breeding warblers at a scale not previously possible for small birds.

    Now, after a year of waiting, I finally had the data in hand.

    How can scientists track small birds?

    Researchers began tracking the movements of migrating birds in the late 1960s using radio telemetry, a technique that requires tagged birds to be physically followed with receiver equipment, sometimes aided by a small aircraft.

    Then, as GPS became more widely available in the early 2000s, satellite tags let researchers accurately track birds remotely. However, GPS transmitters are hefty – even just the battery that powers them is too heavy to be carried by the majority of small migratory songbirds, which often weigh less than a spoonful of sugar.

    To address this problem, researchers developed ingenious little light-level geolocators. These miniature tags minimize weight by containing only a tiny battery, clock, computer chip and light sensor. They store data internally rather than transmitting it – hence my need to recapture tagged birds a year later. Researchers infer a bird’s locations from the timing of sunrise and sunset, which vary predictably across the globe and throughout the year.

    Light-level geolocators have opened new windows into the migrations of small birds, but they also have limitations: Their estimated locations can err by hundreds of miles, and this uncertainty becomes impractically high around the equinoxes, when day length is the same worldwide.

    Two small grey devices sitting on a scientific data sheet.
    Each harness loop of the multisensor geolocator goes around one of the bird’s legs and the tag sits on the bird’s lower back. Stephanie Szarmach

    More recently, researchers developed a newer method of geolocation that uses atmospheric pressure to circumvent these limitations. When a bird ascends during a migratory flight, a barometric geolocator records a steep drop in air pressure, capturing the timing of departure and arrival for every stop along a bird’s migration route. In between flights, the pressure the tag records reflects the elevation at the bird’s position.

    Knowing the bird’s elevation helps to narrow down possible locations, which researchers can then refine further by comparing the change in pressure over time recorded by the tag to global weather data using the computer program GeoPressureR. The unique pattern of rises and falls in pressure acts as a kind of pressure fingerprint for a specific location. Including light and wind data in the model of the bird’s movements can increase precision even more.

    A graph showing atmospheric pressure over time. There are periods of steady pressure, with sharp drops in pressure occurring once per day.
    Atmospheric pressure data collected from a multisensor geolocator carried by a yellow-rumped warbler over five days in spring. When the bird is stationary, such as when resting at a stopover site, the measured pressure remains steady. During migratory flights, when the bird ascends into the sky, the pressure drops and fluctuates dramatically. Stephanie Szarmach

    New tech answers an old question

    We used these new multisensor backpacks to study the migration patterns of myrtle warblers, the northern subspecies of yellow-rumped warbler, which breed across North America’s boreal forest. Myrtle warblers in the northeast are known to winter along the Atlantic and Gulf coasts, but where northwestern populations migrate has been more of a mystery. In 1899, ornithologist Richard McGregor observed myrtle warblers wintering in coastal California, and hypothesized that these birds might breed in Alaska and British Columbia.

    Using these new tags allowed us to test whether warblers breeding in Anchorage, Alaska, actually followed the shortest migration route to winter in California. To our surprise – and probably to McGregor’s as well, if he were still alive – we found that all the birds we tracked traveled east across the boreal forest of Canada before shifting south to the Gulf Coast – certainly not the Pacific Coast route we expected.

    All told, this 6,800-mile (10,900-kilometer) round-trip journey was much longer than we anticipated for a species commonly considered a short- or medium-distance migrant.

    A map showing six geolocator tracks. All tracks start in Anchorage Alaska, before moving eastward across boreal Canada and then after reaching the Great Lakes curving south to end at the Gulf Coast. The predicted wintering area in California is circled.
    Researchers tracked the routes six myrtle warblers flew on their fall migration southeast from Anchorage, Alaska. Figure by Stephanie Szarmach. Warbler illustration by Emily Griffith

    Why take such a circuitous route?

    The long, seemingly suboptimal route taken by our Alaskan myrtle warblers is actually mirrored by some other songbird species tracked from northwestern North America, including blackpoll warblers and Swainson’s thrushes.

    Part of this pattern may relate back to the glaciations that isolated ancient populations of birds millions of years ago. The thinking goes that as these glaciers spread, they squashed the ranges of formerly northern species into the south. Then, as the glaciers slowly receded, the forests gradually reexpanded.

    As more land opened up, birds moved farther north and west into that habitat. Each season some birds went a little farther as the forest available in their northern breeding grounds expanded. For migratory species such as myrtle warblers, the farther north and west their breeding range expanded, the longer the migration to their southern wintering ground became.

    Scientists think migration routes are genetically encoded. The birds’ genes may have traced this expansion front, like Hansel and Gretel putting down breadcrumbs as they journeyed deeper and deeper into the forest. Today’s migration routes still trace that incrementally expanded path, even though the glaciers that shaped the original boundaries are long gone.

    We’re still not sure exactly where the California-wintering myrtle warblers spotted in the 1800s are breeding. We suspect that a small proportion of the Anchorage-breeding birds may migrate to California, but that most of them travel from more distant, unstudied regions of Alaska or Yukon. Future research, potentially tagging myrtle warblers directly on the California wintering grounds, will shed more light on this population and why certain individuals follow a different migration route.

    Barometric geolocation provides new data

    Today, this new geolocation approach improves scientists’ ability to study the migrations of small birds and supports more effective conservation actions across the course of the entire year. More precise location estimates help researchers identify key stopover and wintering sites in need of protection and better understand the threats birds face throughout their journeys. These geolocators also reveal the altitude of migratory flights, which is important for understanding the risk of birds colliding with buildings or wind turbines.

    Because these new tags capture pressure data for every migratory flight, they’re especially valuable for studying migration timing. Researchers have combined data from over 50 bird species to understand at what time of day or night different birds migrate. Now, as a postdoctoral researcher at the Smithsonian Migratory Bird Center, I am integrating multisensor geolocator data with satellite imagery to investigate how arrival and departure decisions throughout a bird’s migratory journey relate to seasonal changes in plant growth.

    Scientists have now tracked dozens of bird species with multisensor geolocators, and that number increases every year. Each mapped migratory journey improves our understanding of how small birds migrate, what factors shape their movements and how best to protect them.

    This article originally appeared on The Conversation. You can read it here.

Explore More Science Stories

Science

Too many males to choose from can overwhelm female frogs – and change who gets to mate

Science

Why do frozen turkeys explode when deep-fried?

Science

Tiny backpacks reveal the secret, sometimes circuitous migrations of small songbirds

Science

Why fall yard care could start with doing less – but not nothing