1 Uncertainty is central to economics

If you rent a car, let’s say in Las Vegas, the rental agent will try to sell you insurance. This additional insurance augments whatever coverage you have from your existing car insurance.

Why does the rental car company offer to sell you insurance? The answer is to make a profit. Their expected cost of the insurance is less than your price. If the insurance seller makes a profit, does this mean that buying insurance is a bad deal for the car renter? Absolutely not. The renter may hate losing big chunks of money so much that even though the expected monetary value of buying insurance is negative, the expected psychological value is positive.

Adam Smith would approve of Las Vegas car renters making a mutually advantageous insurance transaction. The trouble for neoclassical economics comes, however, when the implicitly risk-averse car renter then drives to a Las Vegas casino and begins gambling.

All wagers offered at casinos are profitable, in expectation, for the house. Folk wisdom captures the house’s advantage: “Are you heading to Vegas?” says a neighbor seeing a friend loading luggage into a car. “Yep, I’m looking to win big.” This elicits the retort, “Say hello to my money. I left it there.”

A dollar bet at roulette yields, with common rules, just under 95 cents in expectation. Is it irrational for a person to play roulette at a cost of about one dollar lost for every twenty dollars bet? Not necessarily. Neoclassical economics justifies gambling by inferring that the gambler has “risk-seeking” tastes. The rational gambler gets sufficient joy from considering the possible positive outcomes that the psychological payoff to a money-losing gamble is positive. Voila!

A restaurant offers a hamburger for above cost. An insurance company sells a policy for above expected payout. A casino sells a gamble for more than expected winnings. What is the problem?

The problem, for neoclassical economics, is that people are assumed to have consistent, stable risk attitudes. A person can either buy insurance or gamble. When the same person, however, buys car insurance at the Vegas airport and then proceeds to play roulette, neoclassical assumptions are violated.

People, according to neoclassical economic assumptions, walk through life with their risk-averse or risk-seeking attitudes, making optimal and therefore consistent decisions. Actual human behavior is inconsistent, however, as people are sometimes almost simultaneously risk-seeking and risk-averse.

Why do people make inconsistent and, in some cases, apparently poor, risky decisions? This article explores decisions under uncertainty within the mission of this series (Burnham & Phelan, 2019, 2020a, 2020b, 2020c, 2021a, 2021b, 2021c, 2022a, 2022b):

The Ordinaries column will interpret economic behavior from the perspective of evolutionary biology. From this view of life, the anomalies of behavioral economics will disappear into a coherent biological framework that incorporates elements of neoclassical maximization.

2 Economic views of risky decisions, without biology

2.1 Risk is pervasive

Life is filled with risk. Consider Ötzi the Iceman (Barfield, 1994) who went on a trip more than 5,000 years ago in the Alps near the border between Austria and Italy. Something terrible happened to Ötzi; his mummified body was discovered in 1991 with evidence of conflict that may have led to his death.

Ötzi’s left shoulder contains an arrowhead and there are other wounds on his body. The current consensus is that he bled to death from the arrow, as well as being smashed in the head. Ötzi fell into an ice crack, was frozen, and can be viewed today in the South Tyrol Museum of Archeology in Bolzano.

Economists label any situation with variable outcomes as “decision under uncertainty.” (We will use “risky” throughout this paper.) Ötzi’s life and death definitely met the economic definition of risky. His age at death is estimated at 45 years old, so, while he must have known there was some chance of being killed on the day of his death, the odds were presumably in his favor to survive another day.

Those of us who drive cars face a small, positive chance of death each day. A person who drives 700 miles a week for 50 years has a 2.4% lifetime chance of dying in a car crash (NHTSA, 2022). Every year 5.6% of all deaths occur in some form of accident (cars, ladders, industrial, etc.), so buckle your seatbelts.

Driving with a chance of death is definitely risky. The economics of risky decisions, however, is much broader and encompasses anything that has an uncertain outcome and impacts happiness. If, for example, you dislike having a full trash can, and sometimes the city fails to collect trash, then trash collection is risky.

In fact, nearly every decision is risky from the view of economics. Some of the most important include choice of career, diet & exercise decisions, who to partner with or not, how many children to have, if any, and how to make financial investments.

2.2 Neoclassical economics: expected utility theory (von-Neumann & Morgenstern)

As in other areas, neoclassical economics has a very neat, consistent theory of how humans make risky decisions. The theory is von Neumann-Morgenstern expected utility theory, or simply expected utility theory (von Neumann & Morgenstern, 1944). The central economic notion, which predated expected utility theory by centuries, is that people are assumed to maximize expected happiness, not money.

Should a person buy a lottery ticket with a small chance of winning a billion dollars? In most cases, the cost of the lottery ticket exceeds expected payout. In some cases, with cumulative pots, the price of the ticket can be below the expected payout. Does the expected payout determine optimal behavior? No.

According to expected utility theory, the decision to gamble or not depends on the nature of the gamble (odds and payouts) and an individual’s taste for risk. A risk-averse person may eschew gambles with positive expected value. Conversely, someone who enjoys risk can gamble in Las Vegas and obtain pleasure while losing money.

In fact, faced with a single risky choice, all behaviors are consistent with expected utility theory. A person who refuses to pay a penny for a coin flip to win a trillion dollars is labeled risk-averse, but not irrational. Similarly, a person who gambles his family’s savings away is also not irrational or addicted, but rather risk-seeking.

In summary, neoclassical economics assumes that people make risky decisions in an internally-consistent fashion, based on non-conscious maximization of von Neumman-Morgenstern expected utility. All individual risky decisions are consistent with neoclassical expected utility theory.

2.3 Behavioral economics: prospect Theory (Kahneman & Tversky)

Does expected utility theory describe actual human behavior? No. Almost universally, people are inconsistent in their risky choices in a manner that violates neoclassical economic theory.

In 2022, almost $6 trillion was spent on buying insurance (Insurance Global Market Report, 2022). In that same period, more than half a trillion dollars was lost in legal gambling (Gambling Global Market Report, 2022). Furthermore, gambling revenues are growing rapidly. Most people reading this sentence have both purchased insurance and gambled.

So people spend trillions on insurance and gamble like crazy. Does the combination invalidate expected utility theory, where each person is assumed to be risk-averse or risk-seeking, but not both? To a non-economist, such real world data involving trillions of dollars and billions of people may be persuasive. Behavioral economists, however, prefer laboratory studies using small-stakes or no monetary payoff at all.

Technical violations of neoclassical economic theory constitute behavioral economic “anomalies.” Recall the definition according to Richard Thaler, “An empirical result is anomalous if it is difficult to ‘rationalize,’ or if implausible assumptions are necessary to explain it within the paradigm.” (Thaler, 1987, p. 198).

The seminal work documenting anomalies in risky decision-making is Kahneman & Tversky’s 1979 classic paper ‘Prospect Theory’ (Kahneman and Tversky, 1979). Here is an example of a documented anomaly.

Problem 11:

You have been given $1000. You are now asked to choose between a sure $500 or coin flip for $1000 or no gain.

Problem 12:

You have been given $2000. You are now asked to choose between a sure loss of $500 or coin flip for a loss of $1000 or no loss.

Both of these hypothetical scenarios are the same in terms of payoffs. The choice is between a guaranteed amount and a gamble. And the amounts are identical, although presented in different ways. A person making decisions using expected utility theory, must make the same choice in problem 11 and 12. Actual humans, however, tend to pick the sure outcome in Problem 11 and the lottery in Problem 12.

Thus, the inconsistent choices of people in Kahneman and Tversky’s problems 11 and 12 exemplify a behavioral economic anomaly. There is an enormous behavioral literature documenting divergences between neoclassical economic theory of risky decision-making and actual human behavior.

Kahneman and Tversky summarize their documenting of anomalies as, “The preceding discussion reviewed several empirical effects which appear to invalidate expected utility theory as a descriptive model.” (Kahneman and Tversky, 1979, p. 274).

We agree. Actual humans do not make decisions in a manner that is consistent with expected utility theory.

After behavioral economists document anomalous behavior, their next step is to develop a new theory. In fact, in the same paper where Kahnman and Tversky document anomalies, they propose ‘prospect theory’ as a replacement to expected utility theory.

Prospect theory attempts to make predictions by incorporating perceived empirical regularities in human behavior into a model with more degrees of freedom than expected utility theory. Is prospect theory better than expected utility theory?

Prospect theory was published in 1979. More than thirty years later, a prominent behavioral economist summarized the state of prospect theory, noting that, “It is curious, then, that so many years after the publication of the 1979 paper, there are relatively few well-known and broadly accepted applications of prospect theory in economics.” (Barberis, 2013, p. 173).

In short, prospect theory has failed so far. The author argues, however, that despite decades of failure, the future for prospect theory is bright.

2.4 Economics without biology is lost

In our view, both neoclassical and behavioral economic schools fail to capture the most important aspects of risky decisions. People are wildly inconsistent in their risky decision-making, sometimes appearing risk-averse and at other times risk-seeking. Still other times, people simply make terrible risky decisions. Neither neoclassical expected utility theory nor behavioral prospect theory are accurate summaries of human behavior.

Consider, for example, the case of Detroit police Sergeant Solomon Bell. In 2000, soon after Detroit allowed casinos to open, Sergeant Bell lost thousands of dollars gambling. He then committed suicide by shooting himself in the head. Thousand of gamblers have killed themselves after losing money; what made Sergeant Bell’s suicide more notable is that it occurred inside the casino.

Losing one’s life savings and committing suicide is not a violation of neoclassical economic theory nor a behavioral economic anomaly.

From a neoclassical perspective, both the gambling and the subsequent suicide are assumed to reveal preferences, and each decision is assumed to be optimal. To a behavioral economist, decisions are anomalous only if they contradict neoclassical theory. Because risky gambles and suicide are consistent with neoclassical theory, Sergeant Bell’s gambing and suicide do not constitute anomalies.

In summary, in the area of risky decision-making, economics is lost (see Table 1). The neoclassical model is elegant and internally consistent, but not a useful predictor of human behavior. Behavioral economics has documented divergences between neoclassical economics and actual human behavior, but failed to create a useful replacement theory.

Table 1 Economics and risky decisions without biology, circa 2022
Full size table

3 Biological views on risk

3.1 Our risk attitudes are the product of natural selection

What sort of risk-attitudes are favored by natural selection? At first glance, one might think staying alive is all-important, and thus evolution would favor extreme risk-aversion. The cartoon family in The Croods exemplifies this idea, “New is always bad … Never not be afraid … No one said survival was fun … Curiosity is bad … Fear keeps us alive.”

Risk aversion seems likely to increase lifespan and, thus, might seem to be favored by natural selection. This, however, is not correct.

The currency of selection is genetic replication, not survival. Consider the decidedly risky decision in 1620 to sail on the Mayflower from Europe to North America. Approximately half of the 102 passengers died on the voyage or during the first winter of 1620–1621. Wouldn’t natural selection have favored people who chose not to board the Mayflower?

Consider one passenger in particular, John Howland, who nearly died on the crossing when he fell overboard during a violent Atlantic storm. Howland somehow grasped onto a rope trailing behind the boat, and was pulled back onto the vessel. Subsequently, he survived the first winter in New England.

Starting as an indentured servant, Howland’s economic and physical condition improved; he became assistant to the Governor, and was able to marry. He and his wife, Elizabeth, went on to have ten children and eighty-eight grandchildren. Today, there are estimated to be hundreds of thousands of direct descendants from John Howland, known as “the man who fell overboard” (Philbrick, 2006).

John Howland was a huge genetic risk-taking winner. His risk-taking behavior led to enormous biological success, in the form of high levels of replication of the genes that created his body and brain, and influenced his behavior. Genes that exist today in his descendants.

The point is that appropriate risk-taking – even with a high risk of death – can be favored by natural selection. Migration is risky for any organism. Moving to a new and unfamiliar area, encountering novel organisms and diseases, is almost certain to increase the chance of death. The evolutionary upside, however, is the possibility of enormous reproductive success.

In fact, hundreds of millions of modern Europeans are descended from individuals from a small number of groups of migrants that came from the Middle East and Russia (Racimo et al., 2020). Each of the individuals in these groups likely faced a risky choice: stay in familiar territory or take a chance. (Some migration was, of course, involuntary.)

Because the currency of natural selection is relative reproductive success and not organismic survival, we descended not from risk-averse people, nor from risk-seeking people. Rather, we descended from people who, on average, took good risks. These include many decisions that increased the chance of death.

Even though our world has changed dramatically, we still carry with us the genes of our successful ancestors – genes that persisted because they maximized fitness in the ancestral world. The source of risk-taking troubles (and joys) lies in genes adapted for maximal replication in an environment very different from our own.

How do our genes guide our risky decisions? The answer – back in the ancestral environment as well as in today’s modern environment – is by rewarding the choosing of certain behaviors with pleasure inside our brains. Dopamine – the chief brain chemical that generates pleasure – is involved in risky behavior as it is in many other types of actions.

As a consequence, dopamine is intimately tied to an individual’s taste for risk. One set of academic papers suggests that the molecular structure of an individual’s dopamine receptors influences risk-attitudes (see, for example, Zuckerman & Kuhlman, 2000; Laviola et al., 2003; Dreber et al., 2009; Carpenter et al., 2011; Kohno et al., 2015; Thörnqvist, 2019). We start with the idea and then comment.

The ‘novelty-seeking gene’ has been argued to lie in a person’s dopamine receptor structure. When we take a risk, such as riding a roller coaster, our brains reward us with some dopamine. There is variation between people in the chemical structure – and thus, sensitivity – of the dopamine receptors. In particular, some people have a structure that makes the binding of dopamine to the receptor weaker. For these people, any fixed amount of dopamine leads to less happiness than for people with different types of dopamine receptors.

So, seeking a dopamine reward, people with low-pleasure-receiving versions of dopamine receptors take more risk and seek novelty more than others – physical risks in the form of bungee jumping or drug use, or financial risk in casinos or investment accounts.

To us, the specifics of the novelty-seeking gene idea (i.e., that there is a gene specifying production of the reduced-sensitivity dopamine receptor) are less important than the concept. The concept is simple: our risk-seeking attitudes exist inside our brains. We take risks because doing so leads to the experience of pleasure.

3.2 Ultimate cause of risky decisions

3.2.1 Natural selection produces adaptations

First, we need to define ‘fitness’ and ‘adaptation.’

Biologists use the term ‘fitness’ to summarize evolutionary success. More formally, fitness is the reproductive output of an individual with a particular phenotype, relative to the reproductive output of individuals of the same species with alternative phenotypes.

“Adaptation refers both the the process by which organisms become better matched to their environment and to the specific features that make an organism more fit.” (Phelan, 2018, p 323).

Natural selection favors maximization of fitness, and produces adaptations in the process. At any point in time, in a given environment some organisms have greater fitness than others. If the relevant variation is influenced by underlying genetic variation, then evolution by natural selection will occur.

It is easy to see that morphology – biological form and structure – has been shaped by evolution. The list of physical adaptations is long. For example, porcupine quills make porcupines almost impervious to predation. A seal’s blubber, which enables the seal to survive extreme cold, is an adaptation. Lizards famously can shed their tail quickly – called caudal autonomy – and flee without losing their life.

Coloration, too – the palette of colors and the pattern – can be adaptive. One sex or another of a species is frequently extremely colorful to attract mates. When trying to avoid being eaten, however, camouflage is often the best route. Resting on trees, for example, peppered moths’ wing coloration so closely matches the bark that the moths are nearly invisible to bird predators.

How did this adaptation arise? Even without being there, we know the answer. At some previous point in time, peppered moth coloration did not match so closely the bark of nearby trees. There was variation in coloration among individual moths in the population, influenced by different genetic sequences. Individuals with coloration that more closely matched the tree bark had higher fitness because they were less likely to be eaten by predatory birds. As a result, the population frequency of certain versions of the coloration genes increased, because they produced more cryptic coloration.

In areas where air pollution led to darker soot-covered trees, the coloration of moths in those populations changed, to become darker, matching the trees. Subsequently, when the air pollution was reduced and tree bark again became lighter and more speckled, moth populations changed again, so that they closely matched the tree bark and had reduced predation risk (Cook & Saccheri, 2013; Kettlewell, 1961).

3.2.2 Adaptations can influence behaviors as well as physical traits

Just as there is an extensive set of morphological adaptations, so too are there many well-documented behavioral adaptations. The chest beating of male gorillas signals strength and vigor (Wright et al., 2021). Sea turtle hatchlings swim long distances toward ancestral feeding zones (Lohmann et al., 2017). Birds migrate each winter to richer pastures (Jenni & Schaub, 2003). These are behavioral adaptations that evolved by natural selection to increase survival and reproduction.

Among the many behavioral adaptations produced by natural selection is appropriate risk-taking. What is appropriate? The answer is that, in their ‘natural’ environment, organisms are predicted to, in some situations, make risky decisions – as long as such choices maximize expected fitness. This prediction includes choices that include high-risk of death.

John Howland’s ancestors are more likely than other people to carry the same versions of genes that led to his behavior. Even though few people alive face the same risks as people did in the seventeenth century, the versions of genes that we have inherited are the versions that came from our ancestors. Those genes build our brains and we use those brains to make risky-decisions.

Let’s explore how natural selection has shaped risky decision-making in some other species.

Consider alarm calling in Belding’s ground squirrels. The animals live in groups and are eaten by a variety of predators including birds of prey. When an individual detects a predator, say a hawk, the squirrel has a ‘choice’ to give an alarm call or remain silent. The alarm call increases the chance that the individual sounding the alarm will get eaten. The alarm, however, decreases the chance of other individuals being eaten (Sherman, 1977).

The genetic cost to the alarm is the increased probability of becoming prey. Half of the time, the predation victim is the squirrel making the alarm call. The genetic benefit to sounding the alarm is the reduced chance of other squirrels being eaten, some of whom share common genes with the alarm-sounder.

Belding's ground squirrels have repeatedly faced this situation for countless generations – enough time for natural selection to have produced sophisticated behavioral adaptations. Squirrels sound the alarm in what appears to be an optimal manner by effectively assessing the number and closeness of kin of surrounding animals.

One manifestation of this risky decision-making machinery is that females sound the alarm more often than males. Why? Are females less selfish than males? No.

Females, at maturity, stay in their natal group. Males, on the other hand, leave and travel to new communities. As a result, adult females live in proximity to their mother, father, sisters, offspring, aunts, cousins, younger brothers, etc. Adult males, conversely, live near their daughters and younger male offspring. Therefore, females who sound the alarm are likely to benefit more of their close genetic relatives.

Returning to the economics of risky decision-making. Natural selection favors organisms that make decisions that maximize expected fitness. The biological model is more direct than the neoclassical expected utility maximization, in that the entity being maximized is fitness, not happiness (see, however, Robson, 1996; Robson & Samuelson, 2011, 2019, for explorations of special settings that may favor slightly different risk attitudes). A fitness-maximizing organism would suffer no behavioral inconsistencies akin to those documented by Kahneman and Tversky.

3.3 Proximate causes of risky decisions

How does evolution produce the mechanisms that generate adaptive risky behavior? The answer is that the same process elucidated by Darwin of differential reproduction based on heritable variation in physical traits also shapes physiological and behavioral traits.

We can link morphological and behavioral adaptations to risk by returning to our peppered moths. Cryptic coloration of moths is a morphological adaptation. However, reducing predation by immobility and camouflage isn’t enough to ensure high fitness. A cryptic moth that simply remains hidden forever will find neither a meal nor a mate. Our well-camouflaged moth must, in order to have high fitness, take some risks.

When should the cryptic moth move? What should it eat? Where and how should it find a mate and reproduce? The answers to these questions – based on the actual behaviors we see in natural habitats – reflect a much richer and more complex suite of adaptations than simply coloration. Animals supplement morphological adaptations with behavioral decision-making processes.

Behavioral decision-making processes in animals incorporate risk-perception and risk-assessment. At each point in time, there are benefits to remaining motionless and hidden, but there are also opportunity costs.

In natural populations, we see a wide range of adaptations to reduce predation risk. This is because at each point in time, organisms at risk of predation must make decisions reflecting trade-offs between gaining the food resources they need and reducing their vulnerability to predators (Abrams, 1993; Lima, 1998). And each of these decisions represents an opportunity for adaptations (Lima & Dill, 1990).

The specific behavioral choices that represent risk-assessment opportunities include: when to feed, where to feed, what to consume, and how to consume it (Lima & Dill, 1990). Each of these represents a situation in which natural selection can act to shape organisms’ threat detection and analysis mechanisms and abilities (Blanchard et al., 2011).

Observing animals from natural populations, researchers have identified some consistent outcomes from the evolution of anti-predator strategies. For example, prey sensitivity to predators is strongly influenced by the size, speed, and directness of the predator’s approach (Stankowich & Blumstein, 2005).

A useful metaphor for conceptualizing predation risks is that organisms live within “landscapes of fear” (Laundré, 2001). While the risk of predation may always be non-zero, it still can vary in predictable ways within an animal’s environment. “Fear” can be thought of as an adaptation enabling animals to assign relative risks to activity decisions. And numerous aspects of their environment influence the evolution of such risk assessments.

One literal manifestation of “landscapes of fear” can be seen from satellite images of the barrier reefs as rings around small areas, called patch reefs. The rings – between the reef and the outlying beds of grasses and algae – are mostly barren regions of sand, two to ten meters wide. No algae. No grasses.

Inside of the circle, the patch reefs are structured, high-biomass environments, supporting numerous species of fish and sea urchins. What causes these aquatic “crop circles”? The fish and sea urchins living in the reef forage on anything growing nearby. And while they can consume grasses and other vegetation in a much larger region than their small patch reef, they stay within close proximity to their home reef (Ogden et al., 1973).

Why? Because if the grazing fish and urchins venture into the open water between the reefs, they are at high risk of being eaten by one of the many species of predators. Avoiding that danger zone, they forage close to the perimeter of their patch reef, creating what is known as “grazing halos.”

The message is that feeding behavior represents appropriate risk-taking. In this case, movement within the local zone is a risk worth taking, while movement between zones is not. The animals have evolved to take the risks that pay (or paid for their ancestors).

Observation of the feeding behavior produces a testable hypothesis about the risk-based origin of grazing halos. The hypothesis predicts a reliable relationship between the size of grazing halos and the presence of predator fish. One study tested this hypothesis in regions where fishing is prohibited and therefore more predators were present (Madin et al., 2011). As predicted, when the populations of predator fish are larger, grazing halos are significantly narrower.

The relationship between predator prevalence and the size of the halos is consistent with the hypothesis that with greater risk outside of their patch reef, the grazers opt to stay even closer to the safety it provides. Grazing halos occur in reefs around the world–from the Great Barrier Reef to the Red Sea to the West Indies (Bilodeau, 2021; DiFiore et al., 2019; Randall, 1965).

Analogous landscapes of fear occur in terrestrial habitats, too. The pika (Ochotona princeps), for example, is a small mammal that lives on rocky slopes and forages on a wide range of plant species. Pikas have numerous predators, including weasels, foxes, coyotes, and birds of prey. In their foraging, the pikas restrict their movements to vegetation surrounding a safe refuge, eschewing significant vegetation outside their safe zone (Huntly, 1987).

Across the animal kingdom, we see the effects of natural selection acting on: (1) prey physiology, to better gather information from their environment and increase their sensitivity to predators (Bjaerke et al., 2014; Castorani & Hovel, 2016; Chivers & Smith, 1998; Lima, 1998), (2) general prey behavior, and 3) prey morphology:

Chemicals of fear – Mud crabs, for example, recognize the presence of blue crabs – the chief predator – in their (often murky) habitat by detecting exuded metabolites. The mud crabs responded to the presence of the chemicals trigonelline and homarine by reducing their feeding behavior by 65% – regardless whether the chemicals were in blue crab urine or injected into the environment by the researchers (Poulin et al., 2018).

Landscapes of risk – Mice from three distinct populations were placed in a variety of novel environments – open-field, concentric square field, elevated maze – and evaluated for risk-assessment behaviors (Augustsson & Meyerson, 2004). Two of the populations were commonly-used strains of laboratory mice, genetically modified by hundreds of generations of inbreeding and the absence of natural selection for any predation-risk assessment. Mice from the third population, conversely, were wild-caught mice.

Behavior among the wild mice differed significantly from that of the lab strains. Specifically, the wild mice had reduced activity and much higher avoidance of open areas. Importantly, the exploratory activity of the wild mice was not less than that of the lab strains. Rather, they exhibited more caution and, only after assessing new regions as non-risky, did they enter.

Comforts of safety – Not surprisingly, meta-analyses reveal that animals possessing defensive armor or cryptic coloration, perceive less risk in predatory encounters. These animals wait longer and allow potential predators to get closer to them before they initiate escape behaviors (Stankowich & Blumstein, 2005).

These experiments show a sophisticated risk and return assessment as part of risky decision making in animals.

3.4 Natural selection produces adept risk assessors

Red deer males (stags) frequently face a risky decision. When a female deer is fertile, males compete to gain access to the female and to father the offspring. The decision to compete, however, is risky if the stag is injured. Evolution has produced mechanisms that diminish the costs of this competition.

When four-hundred-pound red deer stags compete for access to females, it is rare for them to actually engage in physical fights. Instead, two males face each other and take turns roaring for several minutes. If one’s roar is substantially greater in magnitude and rate, the other nearly always retreats.

Only if the roars are similar in magnitude and rate does the contest escalate, but still not to a physical fight. Rather, the males conduct a “parallel walk,” where they size each other up. If one of the two males is significantly smaller, he withdraws. Only in a very small percentage of cases does a fight ensue. Such contests might seem to suggest that animals are averse to risk. After all, the fitness payoff is great. Victory in red deer contests leads to high reproductive success, the currency of evolutionary fitness, while losers, and those not competing at all, have none.

Rather than reflecting risk aversion, however, the low levels of physical fighting reveal extremely effective risk perception and assessment. Both roaring and the parallel walk are low-cost behaviors that enable a male to assess the probability of success in an actual fight. Larger, healthier animals have higher mean and maximum rates of roaring. And based on data from situations in which fights occur, mean and maximum roaring rates are extremely highly correlated with fighting ability (r = 0.800 and r = 0.761, respectively; Clutton-Brock & Albon, 1979).

Across a wide range and very large number of animal species – including mammals, reptiles, amphibians, birds, and insects – we see the evolution of risk-assessments. This can lead to the avoidance of lethal combat in situations where the two parties are unevenly matched.

3.5 Some deadly risks are worth taking

Even as we expect effective adaptations enhancing risk perception and assessment in animals, we find that animals often do take life-threatening risks. An evolutionary cost–benefit perspective rationalizes such behavior.

For example, reproductive physiology and physical combat have been investigated in the bowl-and-doily spider (Frontinella pyramitela). In fights between two males, the larger male wins more than 80% of the time (Austad, 1983) – similar to the situation with red deer stags. Consequently, when two male spiders simultaneously approach a female, the smaller male almost always leaves without any physical combat.

An unusual feature of the reproductive biology of these spiders, however, complicates the risk assessment of engaging in combat. The first twenty minutes of mating is a pre-insemination phase. During this time, neither sperm transfer nor fertilization occurs. After the twenty-minute pre-insemination period, sperm induction and fertilization occur for the next 30 min. Once this second phase has begun, there is a strong, positive relationship between the mating duration and the proportion of the female’s approximately 40 eggs that he fertilizes (Austad, 1982).

As a consequence, there is an asymmetry in the fitness value of a female for: 1) a male just approaching the female, and 2) a male that has been mating with a female for twenty minutes – and is about to begin getting a large genetic benefit. Understanding this asymmetry, researchers set up more than 100 encounters between males, with a range of size differences between the males (Austad, 1983).

In these staged encounters, the smaller of the two males (the “resident”) was allowed to mate with a female for 21 min before the second, larger male (the “intruder”) was introduced by the experimenter. The experimental goal was precisely to pick a fight – and it worked.

The payoff and size asymmetry leads to significant and dangerous battles. As is nearly always the case, the larger male still wins the fight ultimately. In approximately 75% of the battles, the resident experiences disabling or fatal injuries. Each additional second of mating, however, results in additional reproductive success, enhancing the payoff of their high-risk decision.

Anthropologist Napoleon Chagnon argues for a similarly life-threatening risky behavior in humans (Chagnon, 1988). Chagnon lived among the Yanomamö, native South Americans who, up until the late twentieth century, survived by hunting, gathering, and small-scale farming. Chagnon reports that a quarter of all men in his study died from violence.

Why did the Yanomamö men in Chagnon’s study risk killing another? Chagnon argues that those who did and survived ended up with more wives and more babies. Chagnon compiled data from 137 men who had killed other men in battle and 243 who had not. The men who had killed others averaged 1.63 wives (polygamy is legal) and 4.91 children, vs. 0.63 wives and 1.59 children for men who had not killed another person.

3.6 Mechanisms influence risk-perception and assessment

Behavioral decision-making processes in animals incorporate risk-perception and risk-assessment. There are both costs and benefits to the choices an animal can make. And while selection favors appropriate risk-taking, this doesn’t mean there will be unambiguously “good” or “bad” choices. Rather, in populations of animals living under natural conditions, we expect – and see – significant variation in the choices individuals make. Physiological and environmental factors influence individual and situations variation in risk-taking behavior.

3.6.1 The impact of hunger on risk assessment

Even for the same individual, an appropriate risk on one day may not be appropriate on another. In an experiment using the soil worm, Caenorhabditis elegans, Ghosh et al. (2016), demonstrated that hunger significantly increases an organism’s willingness to take a risk.

Using specialized chemosensory neurons, C. elegans typically avoid microenvironments with high salt concentration – which can cause desiccation and death – and are attracted to food odor, such as diacetyl (Culotti and Russell, 1978; Solomon et al., 2004). Researchers set up agar plates, with a “safe” central zone, and a region of increased desiccation risk outside of this zone, with a small amount of food odor added to the desiccation risk zone. They then observed the behavior of worms that had been deprived of food for 0, 1, or 5 h.

Among the worms that had not been deprived of food, fewer than 5% left the safe zone and moved toward the food odor. In contrast, among worms deprived of food for 1 h, approximately 50% left the safe zone. And 80% of the worms deprived of food for 5 h left the safe zone and moved toward the food odor!

In addition to this behavioral change, in the food-deprived worms the researchers detected significantly reduced activity in the sensorimotor neurons responsible for perceiving desiccation threat. In essence, hunger changes the neurologic system used to assess the signal of risk, causing the organisms to tolerate greater risk.

3.6.2 The impact of temperature on risk assessment

In ectotherms such as fish, energy expenditures can be significantly influenced not just by hunger, but by temperature as well. Working with damselfish (Pomacentrus chrysurus), researchers investigated how these factors influenced risk-taking (Lienart et al., 2014). Fish exposed to the chemical alarm cues released in response to predators, significantly reduced their foraging. This reduction in foraging occurred regardless of whether the fish had been maintained on a high or low food ration.

When low-food-ration (i.e., hungry) fish were maintained in an environment 3 degrees C above ambient temperature, however, they no longer reduced their foraging behavior in response to the chemical alarm cues. The fish will tolerate greater or lesser predation risk, depending on their physiological state. In this case, the increased energetic demands associated with a higher-temperature environment cause an increase in the predation risk they choose to endure (for similar findings, see Caraco et al., 1990, Houston et al., 1993, and Killen et al., 2011).

Animals – whether mice or crabs or fish or deer or humans – possess a diverse range of adaptations enabling effective risk-perception and risk-assessment. In the next section, we discuss the (remarkably conserved) proximate neurological mechanisms by which organisms carry out these perceptions, assessments, and behavioral decision-making processes.

3.7 Specific brain regions are activated in risky decisions

Within the brain, risky decision-making is characterized by distinct activation of particular brain regions. Increasingly, neuroimaging studies – using humans as well as a wide variety of animals – are identifying the specific brain regions within frontal, parietal, temporal, and occipital cortex that are active in, and often responsible for, each of the facets of risk perception and assessment (Bechara et al., 1994, 1999; Orsini et al., 2015; Román et al., 2019; Talukdar, 2018).

For example, researchers measure increased activity in the parietal cortex in environments with risks of known probabilities (Kiani & Shadlen, 2009; Shadlen & Newsome, 2001; Yang & Shadlen, 2007). Encountering risks with unknown probabilities, however, produces neurally distinct – and stronger – patterns of activity, throughout the ventral striatum, orbitofrontal cortex, midbrain, and the limbic system (Schultz et al., 2008).

Researchers also have been able to identify some specific mechanisms for risk assessment and choice. One study used a clever experimental setup, in conjunction with simultaneous neuroimaging of the subjects’ brains (Kohno et al., 2015). Human research subjects were given the opportunity to pump up (virtual) balloons for cash (see Lejuez et al., 2002). The bursting of the balloon is probabilistic, with each additional pump increasing the chance of popping.

The more the subjects inflate the balloon, the greater their monetary reward. The risk is that, if the subjects pump too much and the balloon bursts, the subject receives no cash. As a consequence, each decision – to make an additional pump or to cash out and keep the accumulated earnings – represents a greater risk and a greater potential reward. Not pumping is the safe option, while pumping is the risky option.

What influenced how many pumps a participant would choose? Prior experience played a significant role. Following trials in which participants burst the balloon, they made fewer pumps than they did following wins. In the trials following losses – from unsuccessful risky decisions – both the amygdala and the hippocampus showed significantly greater activity with each pump of the balloon than their activity following cash-out trials (Kohno et al., 2015). In other words, increased activity levels in these limbic-system structures serve to reduce risky decisions.

These results are consistent with results from different experimental approaches that also suggest a strong role for the amygdala in guiding decision-making, by promoting risk-aversion in reactions to risky situations or following negative outcomes. First, individuals with amygdala damage exhibit less risk-aversion following losses than individuals without such lesions (De Martino et al., 2010). And second, following amygdalectomy, rhesus monkeys exhibit reduced inhibition in novel situations (Mason et al., 2006).

Attempting to evaluate neural activity during situations more similar to those encountered in everyday life, researchers investigated brain activity as subjects categorized traffic situations, from low- to medium- to high-risk (Megías et al., 2018). As in the balloon-pumping experiments, subjects had consistent differences in brain responses as they encountered increasingly risky situations.

Specifically, with situations of increased risk, fMRI measurements indicated increasing activation of a network of limbic structures, including the insula, precentral gyrus, and postcentral gyrus. These results are consistent with numerous additional investigations relating to risk-taking and risk-perception (Christopoulos et al., 2009; Coaster et al., 2011; Ernst et al., 2002; Mohr et al., 2010; Vorhold et al., 2007).

In summary, the brain consistently uses specific regions including the amygdala to learn from risky decisions and modulate future choices.

3.8 The role of dopamine in risk-taking

Dopamine plays an important role in decision-making overall and specifically in making risky choices. The primary sites implicated in risk perception and assessment, including the ventral striatum and the insula, are rich with neurons releasing and responding to the neurotransmitter dopamine. These dopamine neurons play a central role in mediating how humans and other organisms evaluate and integrate risk in decision-making.

Our current model of dopamine neuron functioning is as a “rewards prediction error” (Glimcher, 2011; Sarno et al., 2017). An individual is constantly making estimates of the value of current and future situations. Based on past experience, for example, an animal might expect one unit of food reward for each five units of time. If, however, the animal receives one unit of food in three or four units of time, its prediction was in error. In this case, the world is better than the animal expected. As a consequence, dopamine release occurs, in proportion to the magnitude of the error, and future estimates are altered accordingly (Schultz et al., 1993). Through this process, animals are able to accurately encode the value of events and learn.

The vast majority of behavioral and physiological data are well-aligned with this view of dopamine functioning as a rewards prediction error (see for example, Montague et al., 1996; McClure et al., 2003; O’Doherty et al., 2003; Bayer & Glimcher, 2005; Roesch et al., 2007).

Further, the perspective that dopamine is a learning signal, provides insights into numerous pathological behaviors (Lobo & Kennedy, 2006). Research on pathological gamblers, for example, documents decreased concentration of dopamine (Bergh et al., 1997). Similarly, dopamine agonists increase risk-taking (Quickfall & Suchowersky, 2007; O’Kelley et al., 2012; Riba et al., 2008), often to extreme levels.

3.9 Summary of the biology of risky decisions

Natural selection favors ‘appropriate’ risk taking, where appropriate is defined by evolutionary payoffs. There is an optimal level of risk in any particular situation, trading off, for example, the risk of predation with the benefit of eating.

In a wide-variety of species, studies are consistent with the idea of appropriate risk taking. Deer, spiders, crabs, and humans risk dismemberment and death when the costs justify the expected benefits. Furthermore, organisms use cues to the costs and benefits of risky decisions. Finally, researchers are making substantial progress in identifying the detailed neuroanatomy of risky decisions.

4 Biological economic views of risky decisions

Both neoclassical and behavioral economics have flawed models of risky human decision-making. Neither expected utility theory nor prospect theory are accurate characterizations of human behavior.

Animals in sync with their environment face selective pressure to make risky decisions that maximize fitness. Such fitness-maximizing decisions will have many of the characteristics of expected utility theory including consistency. But if natural selection favors optimization along the lines of expected utility theory, why do we still see people making inconsistent and bad decisions? Biological mechanisms and environmental mismatch are the answer.

Mechanism: Behavior is made by specific physiological machinery activated by particular stimuli (See Ordinaries 7, Burnham and Phelan (2022a) for more on mechanism). Because of this biological, physical implementation, the decision-making processes will never be perfect.

Mismatch: An organism, or a population, can be out of sync with the environment. Mismatch is the topic of Ordinaries 2 (Burnham & Phelan, 2020a) and a central concept of the entire Ordinaries agenda.

Animals in novel environments exhibit costly and inconsistent behaviors (Burnham, 2016; Burnham & Phelan, 2000). Examples from earlier Ordinaries articles include choices that increase the likelihood of human heart disease, sea turtles that suicidally crawl away from the ocean to their death, and birds that ‘impatiently’ choose a small amount of food now and then proceed to cache the food for a distant future.

Humans in modern cities suffer from pervasive mismatch. In almost every aspect of our lives, we make inconsistent and bad decisions. The anomalies of behavioral economics are a subset of the problems caused by humans living with mismatch.

In summary, people make risky decisions using our evolved mechanisms in a mismatched environment. Because of mechanisms and mismatch, our risky decisions are inconsistent and often costly.

Expected utility theory is not a good description of human behavior. Some of the anomalies of behavioral economics are important. However, prospect theory is not a good predictor of human behavior. Because different mechanisms are activated by different, frequently novel, stimuli, people will make risky decisions that are inconsistent and costly.

The biological path forward is detailed understanding of the biological and genetic causes of risky decisions. We have discussed some of the literature in this article. Then, with detailed knowledge, individuals and organizations can develop and deploy variants of the four strategies for improving outcomes (see Ordinaries 4: The causes and cures of self-control issues [Burnham & Phelan, 2020c]).

Human behavior in novel settings is never going to be rationalizable into maximizing some set of preferences, neither neoclassical nor behavioral. As individuals, societies, and scholars, we have to accept and investigate human behavior that was forged by a maximizing process, but is inconsistent and costly in the modern world (Table 2).

Table 2 Economics and risky decisions with biology, circa 2022
Full size table