The Paradox of Helping: Endorsing for Others What We Oppose for Ourselves

A homeless person approaches you during the lunch hour, appealing for a few dollars to buy lunch. You are moved to help, but you have a choice—you could either give the person a portion of your own sandwich to eat, or give them cash. Which would you prefer?

Now, consider a second scenario. Your friend approaches you at lunch, also appealing for a few dollars. Would you give to your friend part of your sandwich or the money?

In both scenarios, the request is the same and the options for how to help are the same. But you may help differently because you believe that these two people will use the aid in different ways. You may doubt the homeless person’s likelihood to spend the cash wisely or even spend it on food at all, whereas you have no doubt that your friend will spend your money appropriately. If this describes your reaction, you’re not alone. Most people prefer to donate their sandwich to the homeless person but their cash to a friend.

With my colleagues Adam Waytz and Nicholas Epley, I have just published research in the Journal of Experimental Psychology: General that explores how people think about giving—and receiving—aid. Our research reveals something fundamental about how humans help. Helpers’ beliefs about how recipients will use aid derive from their perceptions of the recipients’ mind—how rational and reasonable the recipients are, and how much self-control the recipients have. When a recipient seems less rational, like stereotypically a homeless person might, the helper is more concerned that the recipient will misuse the aid.

When people donate a sandwich instead of cash, they limit the recipient’s choice. This kind of choice-limiting help tends to be perceived as “paternalistic.” The recipient can only eat the sandwich to satisfy their hunger, and cannot do much else with it. Conversely, donating cash is less paternalistic and more “agentic” because it allows recipients to choose how to help themselves. For instance, the recipient could satisfy their hunger by buying their own sandwich, but alternatively could buy a bag of candy or a pack of cigarettes.

This example illustrates two kinds of help. Help can be categorized as either more “paternalistic” or more “agentic.” Charities like the RedCross and OxFam tend to provide more paternalistic aid to their recipients (food, shelter, clothes). But other charities, like GiveDirectly, provide more agentic aid (direct cash donations).

People who receive help prefer agentic aid—they want to choose…yet people who give help prefer it to be paternalistic.

Governments make similar decisions. Consider a policy designed to help citizens be healthy. The policy could simply ban certain unhealthy foods (in 2012 New York City’s then-mayor Michael Bloomberg tried exactly that with a divisive “soda ban”). Alternatively, the policy could require that restaurants provide calorie counts so that consumers make informed decisions about whether or not to order something unhealthy. The former policy—banning food—is more paternalistic, and the latter policy—providing calorie information—is more agentic.

So how do policy-makers, charities, and individual helpers decide whether they will be more paternalistic or more agentic when helping others?

To answer this question, we reviewed hundreds of studies about helping, revealing a fascinating paradox. People who receive help prefer it to be agentic—they want to choose. When aid is paternalistic, recipients are more likely to resent the help, less likely to accept it, and less likely to reciprocate (see here, here, and here).

Yet people who give help prefer it to be paternalistic. After all, most of the largest aid organizations in the word (e.g., World Health Organization, Red Cross) and almost all countries provide primarily in-kind aid instead of cash donations. This mentality was summed up in the headline of an op-ed in The New York Times: “Is It Nuts to Give to the Poor Without Strings Attached?”

Why do recipients prefer agentic aid but donors prefer paternalistic aid?

The answer may come from one of the most well-known findings in psychology. People tend to believe that they have greater mental fortitude—are more sophisticated, agentic, and rational—than an average person. We call this robust cognitive bias the “lesser minds problem.” Unlike self-enhancement, the lesser minds bias does not predict that people always perceive themselves more positively than others. It also predicts that people believe that they experience negative emotions, such as shame and guilt, more strongly than others.

Because of this pervasive belief, people think they will use agentic aid wisely and can allot it more appropriately. In contrast, people tend to think that other recipients, with their seemingly weaker mental capacities, would choose less wisely and would be better served by paternalistic aid.

To test this idea, we recruited almost 300 workers from large organizations and asked them to imagine various work policies. For each policy, we provided a more paternalistic version and a more agentic version. For instance, if an organization wanted to increase appropriate attire, they could require a strict dress code or simply suggest one. Next, participants picked which policy would be most effective for one of three groups: themselves, an average employee from another division at work, or an average middle school student. Finally, participants rated their own and others’ capacity to achieve the goal—in this case, dress appropriately.

Of course, participants believed that an average employee would have more capacity to achieve their goals than a middle-schooler. But participants also displayed the lesser minds bias: they believed that they themselves were more capable than either the average employee or child. Of 6 policies that we provided, participants on average selected the paternalistic version 28 percent of the time for themselves, but they selected paternalism 38 percent of the time for another employee and 65 percent of the time for a child.

It’s unsurprising that paternalistic policies seem more appropriate for children than for oneself, but it’s much less obvious that paternalism is more appropriate for other adults than for the self. Yet that is exactly what people seem to endorse.

It gets more interesting. Information that makes a recipient seem mentally weaker can change donors’ giving strategies. In another experiment we provided over 500 donors with information about charity recipients in Kenya and Uganda. Half of the donors were told that some recipients are “smart and ambitious” and the other half were told that some are “illiterate and resigned.” Participants then chose to donate to a paternalistic charity (like the RedCross), an agentic charity (like GiveDirectly), or not donate at all.

The help that individuals in need want—and deserve—may be more like the help you want than you realize.

The donors who read that recipients were “smart” were more likely to give their money to an agentic charity than were donors who read that recipients were “illiterate.” Both groups donated around the same total amount of money, but the messages about mental ability changed which charities were chosen.

It’s possible to change how people view others’ minds; it is also possible to change how people view their own minds. To highlight one’s own mental weakness, we used an American holiday that encourages overindulgence: Thanksgiving dinner. We asked nearly 200 Americans to rate their own ability to resist overeating. Some of the participants were asked before they had eaten while others were asked after the feast. We then asked them to rate how effective different paternalistic policies (such as eliminating desserts in restaurants) would be for them, to encourage healthy eating.

Right after eating Thanksgiving dinner, individuals rated their self-control as worse than it was before dinner. As a result, they thought paternalistic policies would be more effective for themselves.

These experiments demonstrate that subtle experiences can influence beliefs about a person’s capacity to use aid wisely, and this can affect actual helping decisions. This poses a real concern: If people have mistaken assumptions about the minds of those being helped, their good intentions may translate into subpar help.

Consider the case of the charity GiveDirectly, whose mission statement is to “send money directly to the extreme poor.” In 2009 four economists from MIT and Harvard founded this charity privately and tested its efficacy for years before making it public. Their research found that giving money promoted well-being and long-term poverty reduction; in-kind transfers were less effective in those measures. More than 30 published papers have shown consistent results. Nonetheless, many donors have been wary of GiveDirectly’s agentic giving strategy. GiveDirectly’s website reports that one of their most frequently asked questions is whether recipients “blow [the money] on booze.” (Answer: People who received help did not spend any more on “temptation goods,” including alcohol, than they do otherwise.)

Perhaps even more disconcerting, decisions on how to provide aid don’t just impact the immediate recipients. They can also impact future aid decisions, creating a pervasive cycle of paternalistic giving.

In an experiment demonstrating paternalistic cycles, we asked 200 online participants to read about identical refugee families who received aid from the United Nations High Commissioner for Refugees (UNHCR). We told half of the participants that UNHCR decided to provide agentic aid—giving cash directly to the refugees for them to spend on food and shelter. We told the other participants that UNHCR provided paternalistic aid, purchasing the food and shelter for the refugees directly. Participants then rated their beliefs about the refugees’ mental capacities (e.g., “these refugees probably have excellent self-control”).

The participants who believed UNHCR provided agentic aid thought the refugees must have more self-control and agency than those who believed paternalistic aid was provided. These findings are consistent with a very well-known sociological effect called the “Matthew effect,” which suggests that in the absence of perfect knowledge about an individual’s true ability or capacity level, observers will infer that person’s ability based on other sources of information, such as their beliefs. In this case, our participants inferred that recipients must be less agentic and intellectually capable if UNHCR gave them paternalistic aid. These inferences can perpetuate inequality, leading others to make similar inferences.

Our experiments focused on how people choose to help themselves and others. But two big empirical questions remain. First, which type of aid—agentic or paternalistic—is actually more effective? Some existing research suggests the answer may be complicated. For instance, people sometimes reject paternalistic aid, to their own detriment (see here and here). At other times, as the aforementioned data on GiveDirectly highlight, people may choose to provide more paternalistic aid for others even though it can impede their well-being.

Second, when do people perceive aid as agentic or paternalistic? One example of aid that can be ambiguous is a “nudge,” which combines elements of both paternalistic and agentic help. On one hand, nudges may appear paternalistic because they manipulate choice architecture, but on the other hand, many nudges are agentic in that the choice is up to the individual. Our research suggests that how people categorize nudges may affect how they prefer to use them.

Paternalistic policies created with the best of intentions, from food stamps to affirmative action programs, can result in unintended negative consequences for the very recipients they were designed to help. Those who design policy, offer aid, or try to help individuals in need would be wise to think carefully about their assumptions about recipients’ intellect. The help that individuals in need want—and deserve—may be more like the help you want than you realize.