Capuchin monkeys do not show human-like pricing effects

Catapano, Rhia; Buttrick, Nicholas; Widness, Jane; Goldstein, Robin; Santos, Laurie R.

doi:10.3389/fpsyg.2014.01330

ORIGINAL RESEARCH article

Front. Psychol., 02 December 2014

Sec. Decision Neuroscience

Volume 5 - 2014 | https://doi.org/10.3389/fpsyg.2014.01330

Capuchin monkeys do not show human-like pricing effects

$\r\nRhia Catapano$ Rhia Catapano¹

Nicholas Buttrick¹

Jane Widness¹

Robin Goldstein²

Laurie R. Santos¹^*

¹Comparative Cognition Laboratory, Psychology Department, Yale University, New Haven, CT, USA
²Department of Agricultural and Resource Economics, University of California at Davis, Davis, CA, USA

Recent work in judgment and decision-making has shown that a good's price can have irrational effects on people's preferences. People tend to prefer goods that cost more money and assume that such expensive goods will be more effective, even in cases where the price of the good is itself arbitrary. Although much work has documented the existence of these pricing effects, unfortunately little work has addressed where these price effects come from in the first place. Here we use a comparative approach to distinguish between different accounts of this bias and to explore the origins of these effects. Specifically, we test whether brown capuchin monkeys (Cebus apella) are also susceptible to pricing effects within the context of an experimentally trained token economy. Using a capuchin population previously trained in a token market, we explored whether monkeys used price as an indicator of value across four experiments. Although monkeys demonstrated an understanding of which goods had which prices (consistently shifting preferences to cheaper goods when prices were increased), we observed no evidence that such price information affected their valuation of different kinds of goods. These results suggest that human pricing effects may involve more sophisticated human-unique cognitive capacities, such as an understanding of market forces and signaling.

Introduction

Congratulations! You have just won a bottle of wine. You have two options: a 2001 pinot noir that costs $10 or another pinot noir from the same year that costs $50. Which do you choose? You probably chose the more expensive wine. Indeed, when given a choice like this, people tend to pick the most expensive options, whether those options involve alcohol (e.g., Jacoby et al., 1971; Plassmann et al., 2008), meat (Makens, 1964; Bello Acebrón, and Calvo Dopico, 2000), or even cassette players (Dodds and Monroe, 1985, see Rao and Monroe, 1989 for a review). What's more surprising, however, is that our preferences for more expensive goods seem to hold even in cases where the price is arbitrary. For example, Plassmann et al. (2008) allowed participants to sample the same wine when it was labeled as either inexpensive ($5 or $10) or expensive ($45 or $90). Participants reported greater experienced pleasure for wines that were labeled as more expensive, even though what they actually drank was the same in both cases. These results suggest that merely labeling a good as more expensive seems to affect the subjective utility a person experiences from that good.

One might be tempted to write off such findings as the result of a strange demand characteristic; perhaps participants self-report that they enjoy expensive options more in order to signal that they're the kind of person who prefers expensive things. However, some evidence suggests that pricing effects may run deeper than mere demand characteristics. First, Plassmann et al. (2008) found that price affected participants' preferences at the neural level; they found that activation in the medial orbitofrontal cortex (mOFC)—an area of the brain thought to encode the subjective reward utility of different stimuli (see review in Levy and Glimcher, 2012)—was higher when participants thought the wine was expensive than when they thought the wine was cheap. These results suggest that participants actually experienced the wine as tasting better when it was labeled as expensive than when it was labeled as inexpensive. Second, pricing influences how effective people think a good will be. Shiv et al. (2005) allowed participants to pay different prices for energy drinks and observed how well they performed on a set of mental acuity puzzles. People who had paid more for the drink showed greater energy-boosting effects than those who got the drink more cheaply (for a similar finding, see Waber et al., 2008). This result further suggests that pricing effects appear to go beyond mere self-reported differences in preferences; simply changing the price individuals pay for a drink affects not only how well they think it works, but also its actual effectiveness.

Although much work has shown that price affects people's expectations about a good (see also Rao and Monroe, 1989 for a review), less work has explored how such effects emerge in the first place. One possibility is that our expectations concerning price information stem from our experience with how markets operate. As any economics major knows, markets tend to conform to the rules of supply and demand. People prefer products that are particularly good or effective, and thus demand for such effective products should increase. Companies, therefore, will likely end up charging more for products that are especially effective due to the higher demand for such products. This relationship means that better tasting and more effective products are likely to be more expensive. One possibility, then, is that our experience with markets causes us to develop an association between price and value; we come to implicitly assume that expensive products must actually be valuable because otherwise sellers would have to lower their prices. In this way, one could explain the expectations we described above as an extension of our experience with human-like markets. It is also possible that our experience with markets allows us to develop more explicit theories about how markets work—we may come to develop rich beliefs about the connection between price and value based on our own understanding of markets. A second possibility, however, is that our expectations about the connection between price and value have nothing to do with our experience in markets. Instead, our preference for more expensive items may stem from more domain-general processes, ones that are not specific to monetary values or markets.

Are our expectations regarding price and value motivated by a domain-general mechanism or by experience with human-like markets? One way to distinguish between these alternatives is to test a population that does not have rich experience with human markets. Here, we test just such a population—capuchin monkeys (Cebus apella).

Although turning to capuchins may seem at first glance a strange way to test the mechanisms underlying pricing effects in our own species, there are several reasons why this population is well-suited for this question. First, researchers have successfully used capuchins as subjects in economics studies examining the origins of judgment and decision-making biases (Chen et al., 2006; Egan et al., 2007, 2010; Addessi et al., 2008; Lakshminarayanan et al., 2008, 2011). In many of these studies (e.g., Chen et al., 2006), monkeys were trained to trade tokens with a human experimenter for different kinds of food. Monkeys were then allowed to enter a market in which they had to choose between different experimenters who sold different goods at different prices. Using this market method, researchers have observed that capuchins appear to have many of the same strategies and biases as humans (see review in Santos and Chen, 2009). Like humans, capuchins exhibit endowment effects (Lakshminarayanan et al., 2008), loss aversion (Chen et al., 2006), reflection effects (Lakshminarayanan et al., 2011), and choice-induced preference reversals (Egan et al., 2007, 2010). Given that capuchins show many of the same economic biases as humans, it makes sense to examine whether this species shows pricing biases as well.

In addition, recent work suggests that capuchins seem to understand some aspects of price in the context of their experimental market. Chen et al. (2006) tested whether monkeys' choices in their market obeyed the tenets of standard price theory (see Becker, 1962). Capuchins were asked to allocate a set of tokens across two different kinds of food (e.g., apples and grapes) at a cost of one token per food item. Chen and colleagues then introduced a compensated price shift, in which the price of one of the goods dropped (e.g., a subject now received two apples per token rather than one). The researchers then tested whether monkeys switched their consumption after this compensated price shift; did monkeys buy more of the cheaper good after the price change? Chen and colleagues observed that subjects attended to price information, buying more of the cheaper good after the price shift. These results suggest that monkeys' choices in this market obey standard price theory, and thus that monkeys attend to price information in their market in some of the same ways as human consumers do in real markets.

Because capuchins appear to understand price information in token markets, this species can provide a particularly useful test case for distinguishing between the two different accounts of pricing biases described above. Although capuchins seem to understand certain aspects of pricing information in a token economy (e.g., Chen et al., 2006), they lack human-like experience with how price works in real markets. The capuchin token economies differ greatly from those of human participants, particularly with regard to the connection between a good's potential value and its price. If human-like market experiences are indeed necessary for the development of an association between price and value, then capuchins should not show the same kinds of pricing effects as humans do.

In the current studies, we developed a series of experiments to determine whether capuchins show human-like price biases. Capuchin subjects were taught the price of two novel foods in the context of their token economy (see Chen et al., 2006). We then assessed monkeys' preferences for the two goods in the absence of tokens (i.e., during free choice). If monkeys exhibit human-like pricing effects, then they should prefer the more expensive good to the cheaper good when they have a chance to freely choose without paying with their tokens.

Experiment 1 began by teaching capuchins prices for two new foods: differently colored flavored ices. We then allowed monkeys to freely choose between the two colors and tested whether they spontaneously preferred the good that we had told them was more expensive.

Experiment 1

Materials and Methods

We tested seven brown capuchin monkeys (AG, AH, FL, HR, MD, MP, NN) from the population at the Comparative Cognition Laboratory at Yale University (New Haven, CT). All monkeys had participated in a variety of experiments involving making decisions in their token economy (Chen et al., 2006; Lakshminarayanan et al., 2008, 2011). All studies were approved by the Yale Institutional Animal Care and Use Committee.

Testing was conducted in a cubical testing chamber (75 × 75 × 75 cm), which monkeys entered via a sliding door attached to their main large social enclosure. Only one monkey was allowed into the testing area at a time. Monkeys were free to walk into an adjacent section of the enclosure during testing where no other monkeys were present. Two panels on opposite sides of the testing enclosure allowed participants to interact with the experimenters. Each of the panels had two trading holes (5 × 9 cm), spaced such that the participants could reach through one but not both of the openings at the same time (approximately 25 cm apart). During testing, subjects were allowed to choose different foods presented on a small table with a sliding component that was hooked to the outside of the testing enclosure. The sliding component had two trays each positioned to line up with a trading hole when slid up to the chamber, allowing the subject easy access to their contents. Monkeys were presented with 12 tokens (1 inch diameter aluminum disks) which they could use to “purchase” small Flavor-Ice ice chunks (2.5 × 1 cm) of different colors (orange and blue).

Before testing began, each monkey was given one piece of each ice color in order to familiarize them with the taste. The order of color presentation was counterbalanced across monkeys. After familiarization with the different flavors, monkeys began the study. All monkey subjects began on an initial preference phase, followed by a price learning phase, and then a preference assessment phase (see Figure 1 for more details).

FIGURE 1

Figure 1. Description of different experimental conditions and phases.

In the initial preference phase, subjects were given a choice between equal quantities of the blue and orange ice. The goal of this phase was to assess monkeys' initial preference between the two colors of ice. We also wanted to be sure that monkeys did not have strong aversion to one ice color; those monkeys who showed significant preferences were removed from the study. Note that because we did not introduce any tokens during this phase, monkeys had the chance to sample the ices before learning about their prices. During each trial, monkeys were given a choice between the two ices. To do so, the experimenter slid the sliding component away from the testing chamber, placed each piece of ice on its respective tray, and then pushed the sliding component back to allow the subject to choose. The subject then selected one of the two ices; after the subject made its choice, the sliding component was withdrawn and the other ice was removed. We presented subjects with two sessions of 12 trials each, with ice placement counterbalanced across trials. Sessions were run on separate days in order to prevent ice satiation.

After the initial preference phase, subjects moved on to the price learning phase. The goal of this phase was to use the token economy to teach monkeys that one of the two goods could be bought at a discount relative to the other. On each trial, monkeys had a chance to give a token to an experimenter who would return either one piece of one color of ice (the expensive ice), or three pieces of the other color of ice (the cheap ice). Note that the expensive color of ice was priced at three times the price of the cheap color of ice. We chose this difference in price because previous work has shown that monkeys are able to distinguish items that are three times the value of other items (e.g., vanMarle et al., 2006). We conservatively chose which ice was cheaper based on monkeys' initial preferences; monkeys who indicated an initial preference for orange ice (AG, FL, MD) were taught that orange ice was cheap while those that indicated a preference for blue ice (AH, HR, MP, NN) were taught that blue ice was cheap. Testing proceeded as in previous token trading procedures (e.g., Chen et al., 2006; Lakshminarayanan et al., 2008). Specifically, subjects could “purchase” one of the two ices by placing a token into the hand of an experimenter. At the start of each trial, the experimenter placed his hand open to receive the monkey's token. At the same time, he displayed a small dish holding the amount of ice to be traded. Upon receiving a token from the monkey, the experimenter brought the dish up to allow the monkey access. Once the monkey had eaten all the ice, the dish was reloaded, and the next trial began. Each monkey received two sessions of 12 trials with the order of the first color counterbalanced across sessions.

Following the price learning phase, monkeys moved on to the preference assessment phase. The goal of this phase was to see if learning the price of the two ices affected the monkeys' preferences for each of the two colors. To test this, we gave the monkeys a free choice between the two colors (i.e., they could take the ices without having to purchase them using their tokens). The preference assessment phase was identical to the initial preference phase; monkeys again received a free choice between the two colors on the sliding trays. Importantly, the experimenter presenting the trays to the monkey was blind to which ice color had been cheap and which had been expensive, and thus could not influence the monkeys' choices. Assuming monkeys were indifferent between the two ice colors initially, we could test whether monkeys show human-like price effects by examining whether they reliably chose the color shown to be more expensive in the price learning phase when choosing in the preference assessment phase.

Results

We first tested to see if monkeys had an initially strong preference for one of the two colors of ice in the initial preference phase. Two of the monkeys showed a strong and significant initial preference (HG: 12.5%, p = 0.0003, JM: 25.0%, p = 0.02) and thus were dropped from further testing. All other monkeys did not show a significant preference (FL: 29.2%, p < 0.064; AG: 37.5%, p = 0.31; AH: 41.7%, p = 0.54; HR: 50.0%, p = 1.00, MD: 37.5%, p = 0.31; MP: 29.2%, p < 0.064; NN: 41.7%, p = 0.54; all tests exact binomial probability estimates against chance), suggesting that these subjects had no initial preference between the two colors and thus could be used as subjects in the subsequent phases.

Monkeys then went on to learn about the price of the two ices during the price learning phase. Note that in this phase monkeys had no choices—they received equal numbers of trades across the cheap and expensive goods. After learning about the prices, we again tested the monkeys' preferences in the preference assessment phase. Specifically, we compared monkeys' preference for the expensive color in the test condition with their preference for that color before they learned the price information. To do this, we used a repeated measures ANOVA with the color used as expensive as a between subject variable (orange or blue) and time of choice as a within subject variable (before price information and after price information). We observed no main effect of color [F_{(1, 5)} = 0.328, p = 0.591], suggesting that monkeys had no strong preference for one color over another. We also, however, observed no main effect of timing [F_{(1, 5)} = 0.527, p = 0.50]. Monkeys' preferences did not change after learning that one item was more expensive. Non-parametric tests confirmed this finding that there was no effect of price training (Wilcoxon signed rank: Z = 0.94, p = 0.35). We also observed no interaction between color and price [F_{(1, 5)} = 0.004, p = 0.95].

Discussion

In Experiment 1, we taught monkeys that one color of ice was three times more expensive than the other, and explored whether monkeys subsequently preferred the more expensive kind of ice when they had free access to both options. In contrast to what's often been observed in humans (Plassmann et al., 2008), monkeys showed no preference for the more expensive ice. Learning which kind of ice was more expensive in the price learning phase did not seem to affect monkeys' preferences in the preference assessment phase. This result suggests that learning that a food is expensive doesn't seem to make monkeys like it more.

There are, however, a few problems with this study. The first concerns whether monkeys noticed the differential pricing of the two goods. Previous work has shown that this population of capuchins understands the “price” of different goods when such goods are sold in different amounts for a single token (Chen et al., 2006). Experiment 1 assumed that similar presentations of different amounts of food would teach monkeys the specific price of each good. It is possible, though, that monkeys did not attend to this information. In Experiment 2, we add a control condition to test that monkeys attended to the pricing information, testing the monkeys on a price shift condition similar to that used in Chen et al. (2006)'s original study.

A second potential flaw in Experiment 1 is that sampling a food may establish a preference that doesn't change once new price information is learned. In Experiment 1, we first exposed monkeys to both flavors of ice in order to obtain a baseline for their preferences. Unfortunately, it is possible that this experience allowed monkeys to establish a preference that could not be changed by price information. In Experiment 2, we directly explore this possibility by varying whether monkeys had previous experience with the foods whose prices were being manipulated. This manipulation allowed us to examine whether prior experience anchored preferences and contributed to the fact that we did not observe a human-like pricing bias in Experiment 1.