1

In the Eye of the Beholder:

Visual Biases in Package and Portion Size Perceptions

Nailya Ordabayeva a,*

Pierre Chandon b,c

September 16, 2015

a Carroll School of Management, Boston College, 140 Commonwealth Avenue, Chestnut Hill, MA, 02467, USA. Email: nailya.ordabayeva@bc.edu. Tel.: 617-552-2928.

b INSEAD, Boulevard de Constance, 77305 Fontainebleau, France. Email: pierre.chandon@insead.edu. Tel.: +33160724987.

c Institute of Cardiometabolism and Nutrition, 91 Boulevard de l’Hôpital, Paris 75013, France.

* Corresponding author.

2

In the Eye of the Beholder:

Visual Biases in Package and Portion Size Perceptions

Abstract

As the sizes of food packages and portions have changed rapidly over the past decades, it has

become crucial to understand how consumers perceive and respond to changes in size. Existing

evidence suggests that consumers make errors when visually estimating package and portion

sizes, and these errors significantly influence subsequent food choices and intake. We outline

four visual biases (arising from the underestimation of increasing portion sizes, the

dimensionality of the portion size change, labeling effects, and consumer affect) that shape

consumers’ perceptions of package and portion sizes. We discuss the causes of these biases,

review their impact on food consumption decisions, and suggest concrete strategies to reduce

them and to promote healthier eating. We conclude with a discussion of important theoretical

and practical issues that should be addressed in the future.

Keywords: Portion size; Size estimation; Visual perception; Visual bias

3

In the Eye of the Beholder:

Visual Biases in Package and Portion Size Perceptions

Highlights

1. Supersized portions appear smaller than they really are because of the underestimation

bias.

2. Supersized portions are underestimated more strongly when all three dimensions (L, W,

H) change.

3. Small and healthy-sounding labels and brands reduce perceived calorie content and

portion size.

4. Portion size perceptions are independent of body mass but are more accurate for

conflicted eaters.

5. Biased size perceptions influence portion size preferences and choices but remedies exist.

4

In the Eye of the Beholder:

Visual Biases in Package and Portion Size Perceptions

When making decisions about food, consumers tend to rely more on judgments of food

quality than food quantity. For example, the vast majority of consumers think that to lose weight,

it is more important to monitor what they eat than how much they eat (Collins, 1996; Rozin,

Ashmore, & Markwith, 1996). This focus on quality over quantity is reflected in many dietary

guidelines and weight-loss programs which prioritize eliminating certain types of foods or

nutrients (e.g., sodas, carbohydrates) over regulating the total food intake (Thompson &

Veneman, 2005). As such, consumers may expect to gain more weight from eating very small

portions of a food perceived as “unhealthy” (e.g., one mini-Snickers® bar containing 47 kcal)

than from eating a very large quantity of “healthy” food (e.g., one cup of low-fat cottage cheese,

three carrots and three pears, with a combined calorie count of 569 kcal) (Oakes, 2005).

In the meantime, the sizes of food packages and portions have changed dramatically (Nestle,

2003; Rolls, Morris, & Roe, 2002). Portions grew by 60% for salty snacks and 52% for soft

drinks in the course of just 20 years (Nielsen & Popkin, 2003). Due to public concerns about the

negative implications of supersized portions for consumer health (Ledikwe, Ello-Martin, &

Rolls, 2005; Young & Nestle, 2003), some marketers have attempted to downsize their products,

but with mixed results (Deutsch, 2007). A few downsizing attempts have successfully attracted

health and budget-conscious consumers (e.g., T. G. I. Friday’s “Right Portion Right Price” menu,

Horovitz, 2007). However, other downsizing attempts have tried to pass all the cost to consumers

(e.g., by charging the same price for a smaller size) or to conceal the size reduction through

product packaging (e.g., by replacing some of the product in a package with air). These tactics

5

have drawn strong criticism for deceiving unsuspecting consumers who typically fail to check

quantity information (Grynbaum, 2014).

In view of these trends, it has become crucial to understand how consumers perceive and

respond to changes in package and portion size. In this article, we review four systematic visual

biases that drive consumers’ perceptions of package and portion size, show how these biases

influence food consumption decisions, and suggest how they can be reduced. We conclude with

a discussion of potential directions for future research.

FOUR TYPES OF BIASES AND THEIR REMEDIES

Although information about food quantity is increasingly easy to find, including in

restaurants, consumers rarely consult quantity labels (Wansink & Chandon, 2014). Instead,

consumers tend to base their food purchase and consumption decisions on instant visual

impressions of package and portion size. This is because they expect the package to be a reliable

proxy for the amount of food inside (Lennard, Mitchell, et al., 2001), and because some people

find quantity information difficult to process, especially when it is presented in non-metric units

(Viswanathan, Rosa, & Harris, 2005).

Unfortunately, visual perception is not a reliable indicator of food portion or package size

because of four types of visual biases, which pertain to the underestimation of package or portion

size, dimensionality effects, labeling and affective biases. Below we outline the consequences of

these biases for consumption decisions and discuss the effectiveness of various debiasing

strategies.

6

Underestimation biases and their remedies

The first bias in perceptions of package and portion size is the underestimation of the

magnitude of the increase in the actual size of a package or a portion, whereby the perceived size

grows more slowly than the actual size. Essentially, when consumers encounter a new supersized

product (e.g., a new extra-large can of soda), they underestimate how much larger it is compared

to the existing smaller size that they remember from prior purchase or that is displayed next to

the supersized product on store shelves or on restaurant counters..

A stream of literature in psychophysics has established that the underestimation bias arises

because people’s visual perceptions of the size of physical objects follow an inelastic power

function of actual size, as captured by the following mathematical expression (Stevens, 1986):

ESTSIZE = a × ACTSIZE b,

where ESTSIZE is estimated object size, ACTSIZE is actual object size, a is a constant term and

b is the power exponent denoting the sensitivity of size estimations to changes in actual size.

In the context of food packages and portions, marketing studies have demonstrated that the

power exponent b of consumers’ size perceptions typically ranges between .5 and .8 (Krishna,

2007, 2012), with values close to 1 observed only for one-dimensional figures such as lines

which are rarely encountered in the food domain. This means that the sensitivity of consumers’

size estimations to actual size diminishes as the packages and portions grow bigger, resulting in

the underestimation of large sizes. In other words, consumers become increasingly desensitized

to package and portion size as packages and portions grow bigger, with the result that they

choose and consume larger portions without realizing just how large these portions really are.

7

To test these predictions, Chandon and Wansink (2007a) asked the customers of fast food

restaurants in three US cities to estimate the number of calories contained in meals that they had

purchased. The results showed that consumers underestimated the size of their meals (the

average estimated size was 546 Kcal compared to the actual 744 Kcal, a 27% underestimation);

more so for large meals (estimated size was 687 Kcal vs. actual 1144 Kcal, a 40%

underestimation) than small meals (estimated size was 433 Kcal vs. actual 484 Kcal, an 11%

underestimation). This underestimation bias was replicated for round and square shapes (e.g.,

pizzas, Krider, Raghubir, & Krishna, 2001), when quantity increased or decreased (Chandon &

Ordabayeva, 2009), as well as for food kept in the pantry (Chandon & Wansink, 2006).

Importantly, the prevalence and the magnitude of the underestimation bias did not depend on

body mass or people’s knowledge about nutrition and portion size. The same authors (Chandon

& Wansink, 2007a; Wansink & Chandon, 2006a) reported a similar pattern of underestimation

among normal-weight and overweight individuals, among people with a high or low interest in

nutrition, and even among trained dieticians. These findings are consistent with the prevailing

view that visual biases are hardwired (Raghubir, 2007), suggesting that information-, attention-,

and motivation-based strategies directed at debiasing size perceptions have limited effects.

Accordingly, providing information about the underestimation bias does not improve the

accuracy of consumers’ size perceptions (Chandon & Wansink, 2007a; Ordabayeva & Chandon,

2013). Drawing attention to the product does not improve estimation accuracy (Folkes & Matta,

2004), neither does consumers’ motivation to produce accurate estimates, be it inherent or

induced through financial incentives (Ordabayeva & Chandon, 2013; Raghubir, 2007).

Instead, studies have found that piecemeal estimation, consisting of breaking up a large meal

into its individual components (e.g., the main dish, the side dish, and the beverage) is an

8

effective way to reduce the underestimation bias. This is because the individual components of a

meal have a smaller size than the total meal and therefore the perception of their size is more

accurate. Thus piecemeal estimation of food portions improves consumers’ perceptions of total

meal size and, as a result, weakens preferences for large meals, even among dieticians (Chandon

& Wansink, 2007a). To illustrate, in one study dieticians were asked to estimate the sizes of

small, medium and large fast-food meals containing a sandwich, chips, and a drink. They then

indicated which of the three meals they would choose to consume afterwards. Those dieticians

who estimated the size of individual meal components had more accurate size estimates and were

more likely to choose a small meal than those who estimated the size of the full meal directly.

Figure 1 summarizes these findings.

---Insert Figure 1 about here---

Helping people realize just how large today’s supersized portions really are may also

enhance the sensory pleasure that people derive from the meal. Indeed, studies have shown that

supersized portions often yield lower sensory pleasure because of sensory-specific satiation

(Garbinsky, Morewedge, & Shiv, 2014; Cornil & Chandon, 2015a, 2015b).

Dimensionality biases and their remedies

The second bias in consumers’ size perceptions arises from the shape of the change in

package or portion size, notably the dimensionality (whether the package or portion changes in

one, two or three dimensions, and whether the dimensions change in the same or in opposite

directions). In short, the underestimation bias is lower when a package or a portion changes in

one dimension (e.g., when a tube of Lay’s Stax® potato chips increases only in height) than in

9

multiple dimensions (e.g., when a bag of Lay’s potato chips increases in height, width and

length). Size perception is even less accurate when some dimensions change in opposite

directions, such as when the base of a package decreases while the height increases (elongation).

Reviewing the literature in psychophysics, Krishna (2007) reported that the elasticity of size

estimations (i.e. the sensitivity of consumers’ size estimations to changes in actual size, as

denoted by exponent b in the power function in the previous section) depends on the

dimensionality of an object. Specifically, the elasticity of size estimates is higher for one-

dimensional objects such as lines (b = 1.0) than for two-dimensional objects such as squares (b is

between .7 and .8) or three-dimensional objects such as cubes (b = .6). A later study showed that

this was not driven by the nature of the object itself, but by the dimensionality of the size change

(Chandon & Ordabayeva, 2009). It is easier for individuals to track volume changes in packages

or portions when the change happens in a single dimension (e.g., when a box of candy doubles in

size by doubling its length) than when the change is compounded across multiple dimensions

(e.g., when the box doubles in size by increasing each dimension by 26%, since

26%×26%×26%=100%). For example, in one study Chandon and Ordabayeva (2009) asked

people to estimate the sizes of six packages of detergent and wool that doubled in size either

through a change in one dimension (the display of detergent increased in width, the display of

wool increased in length) or a change in three dimensions (the detergent box increased in length,

width and height, the wool increased in diameter). As shown in Figure 2, the sensitivity (and

accuracy) of size estimations was significantly lower for three-dimensional size increases (b =

.68) than for one-dimensional size increases (b = .93).

---Insert Figure 2 about here---

10

The dimensionality bias also shapes consumers’ price expectations, product choices and

usage behavior. Consumers expect greater quantity discounts (by up to 57%) for packages

supersized along three dimensions, exhibit a weaker preference for supersized meals and

products (by up to 32%) when the supersizing occurs along three dimensions, and are more

likely to over-pour liquids (by up to 19%) into a conical container (which increases in height and

diameter) than a cylindrical one (which only increases in height).

The dimensionality bias does not occur because people have more difficulty tracking changes

in multiple dimensions than in one; it arises because people have difficulty compounding the

changes along all three dimensions. In fact, other studies have found that size perceptions behave

as if people use an additive rather than a multiplicative rule to combine the changes along

individual dimensions into a single estimate of volume change (Ordabayeva & Chandon, 2013).

According to the additive change heuristic, a package which doubles in size through a

proportionate (26%) increase in all dimensions is perceived to have grown by just 78% (because

26%+26%+26%=78%).

One of the implications of this linear (additive) heuristic is that the dimensionality bias can

be reduced by linearizing the change in volume from multiple spatial dimensions to one.

Accordingly, if the goal is to boost consumers’ preference and willingness-to-pay for supersized

packages and meals, while simultaneously reducing overconsumption, then packages and meals

that increase in only one dimension should be preferred and endorsed. If, however, the goal is to

boost consumers’ acceptance of downsized packages and portions, then a three-dimensional

change in volume should be preferred because it is less likely to be detected than a one-

dimensional change. For example, in one study, the acceptance of a downsized soda pack

11

increased by up to 21% when the pack shrank in height and diameter as opposed to just height

(Chandon & Ordabayeva, 2009).

Moreover, the use of the additive change heuristic implies that product downsizing can go

completely unnoticed if the product is downsized through elongation. Prior research found that

elongated packages appear to be larger than short, wide packages of the same volume (Raghubir

& Krishna, 1999; Wansink & van Ittersum, 2003). A more recent study found that the additive

change heuristic can explain the elongation effect by suggesting that people fail to compound the

reduction in the two dimensions of the base of the product (Ordabayeva & Chandon, 2013).

Furthermore, the additive change model can actually recommend specific changes in the

individual package or portion dimensions that can potentially completely hide product

downsizing. Ordabayeva and Chandon (2013) verified these model predictions in one study, in

which they asked participants to estimate the sizes of objects that decreased by 8%, 16% and

24% in size from the reference through a reduction in height only (1D downsizing) or through

elongated downsizing (achieved by elongating the object’s height and shrinking its base, or by

elongating its length and shrinking its height and width) which the additive change model

predicted would not be perceived. When people relied only on visual estimation of product size,

they underestimated the magnitude of the downsizing that occurred in 1D (a 24% reduction in

actual size was perceived as a 20% reduction), and hardly noticed the downsizing that occurred

through elongation (a 24% actual size reduction was perceived as only a 2% reduction).

Allowing people to feel the weight of the product (visual and haptic estimation) helped to

eliminate the errors in estimating elongated downsizing, but only partly (a 24% actual size

reduction was perceived as a 9% reduction). Elongated downsizing is so potent that, in a

different study, buyers of a known food brand were significantly more likely to accept product

12

downsizing when the package was downsized through elongation (using less dramatic changes in

shape so as to respect design and shelving constraints) (Ordabayeva & Chandon, 2013).

Other studies found that transparent packaging improved volume estimation by facilitating

the individual’s ability to monitor product quantity inside the package as the product was being

consumed (Deng & Srinivasan, 2013). However, this positive effect of transparent packages on

consumers’ ability to monitor product (and consumption) volume may be trumped by the

detrimental effect of transparent packages on self-control due to the greater salience of tempting

foods inside the packaging. Consumers’ desire to regulate food intake and a product’s inherent

ability to facilitate (or jeopardize) this goal may thus interfere with perceptions of package and

portion size. We consider these interactions in a later section on affective biases in package and

portion size perceptions.

Labeling biases and their remedies

The third bias that drives consumers’ package and portion size perceptions arises from food

size labels and front-of-pack labels. While various food labels have been linked to perceptions of

taste (e.g., Levin & Gaeth, 1988; Raghunathan, Naylor, & Hoyer, 2006; Wansink & Park, 2002),

the focus here is on size labels that have been shown to drive perceptions of package and portion

size beyond simple considerations of taste.

First, labeling the sizes of food packages or portions as “small,” “medium,” or “large”

significantly affects size perceptions. In a series of studies, Aydınoğlu and Krishna (2011) found

that consumers perceived the portions of snack foods (pretzels, nuts, small sandwiches and

cookies) to be smaller when they were labeled “small” rather than “medium,” or “medium”

13

rather than “large.” This, again, encourages consumers to eat more while thinking that they are

eating less, an important issue because the use of size labels is not standardized and varies

significantly across stores and restaurants (Young & Nestle, 1998). For example, the “medium”

size of popcorn sold at AMC theatres contains fewer calories (680 Kcal) than the “small” portion

of popcorn available at Regal Cinemas (800 Kcal). Similarly, a medium serving of French fries

available at McDonald’s contains virtually the same amount of fries (117 g) as a small serving

available at Burger King (116 g).

Health and nutrition-related labels also significantly bias calorie perceptions, which are

typically construed as a measure of meal size. For example, labeling a food “low-fat” creates an

inference that the food contains fewer calories (Wansink & Chandon, 2006b), hence consumers

eat more of it because they think a larger serving size is appropriate and feel less guilty during

consumption. The effect of “low-fat” labeling is particularly pronounced for hedonic foods that

induce guilt during consumption, and among overweight individuals who are more susceptible to

guilt-reduction cues. Providing information about the actual number of servings contained in a

portion can curb the effect of “low-fat” labeling on food intake, but only among normal-weight

individuals for whom the serving-size effect of “low-fat” labels is dominant, and not among

overweight individuals for whom the guilt-reduction effect of “low-fat” labels is dominant.

Highlighting the presence of healthy ingredients in a meal also biases consumers’

perceptions of the calories it contains. When Chernev and Gal (2010) asked participants to

estimate the calorie content of a hamburger with or without a side of broccoli, they found that

consumers mentally averaged (instead of adding) the calorie content of a hamburger and broccoli

to compute the size of the combined meal. This resulted in lower (rather than higher) calorie

estimations for the hamburger with broccoli (665 Kcal) than for the one without (761 Kcal). The

14

averaging bias is particularly pronounced when attention is drawn to the healthiness of meal

ingredients, for example, by asking consumers to compare the healthiness of the individual

components of a combined meal prior to eliciting their estimations of meal size. The averaging

bias is mitigated when attention is diverted away from the healthiness of individual ingredients to

the size of the ingredients, for example, by asking participants to compare the size of individual

meal components prior to eliciting their estimations of meal size.

Similar to highlighting the healthiness of meal ingredients, communicating the healthy

positioning of the food or restaurant brand significantly biases calorie perceptions. A series of

studies on perceptions of meals served at McDonald’s (which is perceived as relatively

unhealthy) and Subway (which advertises its healthy positioning) found that consumers, on

average, thought that meals contained fewer calories when served at Subway than at McDonald’s

(Chandon & Wansink, 2007b). Consumers were also less sensitive to changes in actual portion

sizes served at Subway than at McDonald’s, as shown in Figure 3. For example, a 1000 Kcal

meal would, on average, be perceived to contain 744 Kcal at McDonald’s and only 585 Kcal at

Subway. A follow-up study showed that this “health halo” created by healthy positioning led

diners to eat more at Subway than at McDonald’s (1011 Kcal vs. 648 Kcal, respectively), while

thinking that they ate less (487 Kcal vs. 600 Kcal, respectively). As shown in Figure 4, an

effective strategy to reduce this labeling bias is to encourage consumers to consider a

counterfactual (“consider-the-opposite” condition); specifically, how the healthy positioning of a

restaurant may not generalize to the healthiness of all the food items consumed at the restaurant

(Chandon & Wansink, 2007b). Figures 3 and 4 summarize these results.

---Insert Figures 3 and 4 about here---

15

Finally, the design of the food label itself can influence consumers’ perceptions of package

size. Deng and Khan (2009) reported that a cookie package was perceived to be heavier when the

image of cookies was located on the right (vs. left), on the bottom (vs. top), or on the bottom-

right (vs. top-left) of the package facing. This image location bias could be reduced if the

package was placed next to packs with a similar product location displayed on the store shelf

(such that a package with a product displayed in the “heavy” or “light” location would be

assimilated to the surrounding packs that displayed products in the same “heavy” or “light”

location, respectively). These findings suggest that pictorial information provided on food labels

should be considered in addition to the verbal information when predicting the effects of labels

on package and portion size perceptions. Specifically, consumers may eat more out of packages

that display the product in “light” (vs. “heavy”) locations because they underestimate how much

product the package actually contains. This means that strategically managing the location of the

product display on the package facing and the location of the product in the supermarket aisle in

ways that reduce the underestimation of package content may curb overconsumption.

Affective biases and their remedies

A final form of bias in package and portion size perceptions arises from consumers’

affective reactions to food. Early studies in social psychology found that a desire for an object

could distort perceptions of object size. For example, cigarettes appeared to be larger to smokers

than to non-smokers (Brendl, Markman, & Messner, 2003); water appeared to be closer to thirsty

than to non-thirsty individuals (Balcetis & Dunning, 2010); a muffin appeared to be larger to

dieters than to non-dieters (van Koningsbruggen, Stroebe, & Aarts, 2011).

16

Recent studies have shown that this misperception of food size arises not just from the

greater desire that individuals experience towards hedonic foods, but from the broader conflict

that they experience between their desire for hedonic foods and the perceived health risk of these

foods (Cornil, Ordabayeva, Kaiser, Weber, & Chandon, 2014). Emotional conflict or

ambivalence towards food actually improves (rather than reduces) consumers’ sensitivity to

increases in package and portion size and thus improves estimation accuracy. This is probably

because conflict draws greater attention to portions of hedonic foods and increases motivation to

accurately estimate these portions.

Cornil and colleagues (2014) found that individuals who inherently felt conflicted towards

hedonic foods – for example, people who regulated their food intake or those who struggled with

self-control (Papies, Stroebe, & Aarts, 2008; Scott, Nowlis, Mandel, & Morales, 2008) – had

more accurate perceptions of portion sizes of hedonic foods. When the researchers asked a group

of restrained eaters (who typically experience a strong desire for hedonic foods) and unrestrained

eaters at an urban gym to estimate six portions of potato chips, they found that restrained eaters’

estimations of “regular” chips were more accurate than their estimations of “low-fat” chips and

than estimations of “regular” and “low-fat” chips made by unrestrained eaters (see Figure 5).

---Insert Figure 5 about here---

More importantly, Cornil and colleagues (2014) found that the accuracy of consumers’

portion size perceptions could be improved with manipulations that induced emotional conflict

towards food, i.e. interventions that simultaneously boosted the desire for food and perceptions

of the food’s unhealthiness. In one study this was achieved by asking participants to sample a

small amount of potato chips (which induced a desire for chips, Wadhwa, Shiv, & Nowlis, 2008)

and by labeling the chips as “regular” (which induced perceptions of the chips’ unhealthiness).

17

While these findings have made significant strides toward uncovering the potential role of

affect in portion size perceptions, work in this area is still underway. Clearly, it is important for

future studies to further investigate the role of specific emotions and motivations in size

perception biases. We elaborate on this point in the General Discussion.

GENERAL DISCUSSION

This article has reviewed four prominent factors that bias consumers’ perceptions of package

and portion size. We have also outlined the strategies that can effectively mitigate these biases

and improve the accuracy of consumers’ package and portion size perceptions. Table 1

summarizes the key results.

---Insert Table 1 about here ---

While the studies reviewed in the article cover significant ground in describing and

explaining biases in package and portion size perceptions, a number of important issues remain

unexplored. First, much of the research covered here has studied the effects of cognitive and

affective factors on size perceptions separately, while it is possible that these factors interact. In

the same way that affect influences individuals’ cognitive strategies and information processing

in domains such as health risk perception and product evaluation (Keller, Lipkus, & Rimer,

2002; Pham, 2007), affective responses to food may influence the cognitive strategies (e.g., the

size computation heuristics) employed to estimate package and portion size. Future research

should explore these possibilities in order to devise better strategies to improve size perceptions.

Future research should also examine the effectiveness of novel strategies to reduce size

estimation biases. Whereas prior work has shown that small portions are associated with both

18

higher estimation accuracy and greater consumption pleasure and flavor perceptions (Cornil &

Chandon, 2015a, 2015b; Morewedge, Huh, & Vosgerau, 2010), it would be interesting to test

whether focusing on the hedonic pleasure or the rich flavor of a meal improves consumers’

perceptual sensitivity to changes in food size. It could also be useful to identify tactics that

enhance the predicted and actual pleasure derived from food consumption. One possibility may

be to encourage the use of consumption rituals (e.g., simple gestures such deep breathing or

closing eyes, Vohs, Wang, Gino, & Norton, 2013), which have been shown to enhance the

enjoyment of hedonic foods such as chocolate as well as utilitarian foods such as carrots.

Similarly, it would be valuable to test the effectiveness of additional debiasing strategies that

can be easily implemented in the marketplace. One such tactic may be drawing attention to

foods’ unit price in the retail environment. While the display of unit price is not uniformly

mandated, many retailers provide unit price information in their stores voluntarily. Prior studies

have suggested that perceptions of package size and unit price are linked (Nason & Della Bitta,

1983; Wansink, 1996), suggesting that available unit price information may be used to infer

package size. However, some consumers pay little attention to unit price information (Vanhuele

& Drèze, 2002), and even when they do, they often misunderstand price and quantity

information. For example, Mohan, Chandon, and Riis (2015) reported that most people, even

highly educated ones, believed that 50% more product for free is as valuable as a 50% price

reduction, whereas, in fact, 50% extra product volume reduces the unit price by only 33%.

Although providing unit price information could reduce such errors, a significant proportion of

consumers fail to recognize that a given percentage cost discount is more valuable than a

quantity bonus of the same nominal magnitude. Therefore, finding ways to make unit price

19

information more prominent and influential in the size estimation process would help consumers

make healthier in-store decisions.

More generally, price and cost-related issues have received limited attention in the size

perception literature. While biased size estimates influence consumers’ price expectations

(Chandon & Ordabayeva, 2009), less is known about how price informs the size estimation

process. In the meantime, changes in package and portion size are typically accompanied by

changes in price (Mohan, Chandon, & Riis, 2015). It is important to understand how consumers

combine and trade off package size information with price information to make food decisions.

Much of the research to date has focused on the effects of product-related factors on package

and portion size perceptions, leaving the effects of non-food-related, environmental factors

largely untapped. Yet environmental cues play a significant role in how consumers perceive and

consume foods (Chandon & Wansink, 2012), and meal size perception is no exception. One

study revealed that perceptions of meal size served on a plate were subject to the Delboeuf

illusion: the same portion looks smaller on a large plate than on a smaller plate (Van Ittersum &

Wansink, 2012). Other studies have shown that various sensory modalities (for example, sound

pitch, food color) influence a host of food perceptions (Spence & Piqueras-Fiszman, 2014), and

may therefore influence size perception as well.

Similarly, a more refined understanding of the influence of the sensory characteristics of the

food itself on package and portion size perceptions is needed. For example, a recent study

reported that a food’s texture influenced perceptions of its calorie content: foods with soft

textures were perceived to be more caloric than foods with hard textures (Biswas, Szocs,

Krishna, & Lehmann, 2014). It would be interesting to study how the experience of the food

through other senses (e.g., smell, taste) influences visual estimations of food size. In this context,

20

examining the effect on size perceptions and consumption of novel food processing techniques

such as the infusion of air into food (e.g., ice-cream) through a slow-churning process will be

practically relevant.

Finally, the current understanding of how visual biases influence behaviors other than size

estimation including food choice, portion size choice, ingestion, satiation and product usage

remains relatively limited. Future research should more comprehensively delineate the

behavioral outcomes of visual biases across lab and real-world settings.

In addressing these various questions, future studies will need to go beyond the traditional

methods to test the practical relevance and implications of size perception biases and solutions in

a natural consumption setting. This will entail expanding beyond the laboratory to environments

in which food consumption actually takes place, from undergraduate student samples to more

representative and larger populations, including vulnerable groups (e.g., children, individuals

who are actively trying to lose weight or those who are trying to manage weight-related health

issues), from estimations of food that is about to be consumed or has just been consumed to

perceptions of food that is prepared (cooked) for oneself and others, food that is stored or

stockpiled, and food that is wasted. Understanding how one-shot judgments of food quantity and

quality (which have been the subject of most of the existing studies) translate to repeat behaviors

and habits, and how interventions that improve food perceptions suggested in the literature

endure over time are another avenue to be explored. Such approaches would enable a long-term

perspective to emerge about the roots and remedies of size perception biases and thus allow us to

address crucial questions about the effectiveness and sustainability of current food packaging and

portion strategies for consumer health and wellbeing.

21

REFERENCES

Aydınoğlu, N., & Krishna, A. (2011). Guiltless gluttony: The asymmetric effect of size labels on

size perceptions and consumption. Journal of Consumer Research, 37 (6), 1095-1112.

Balcetis, E., & Dunning, D. (2010). Wishful seeing: More desired objects are seen as closer.

Psychological Science, 21, 147–152.

Biswas, D., Szocs, C., Krishna, A., & Lehmann, D. R. (2014). Something to chew on: The

effects of oral haptics on mastication, orosensory perception, and calorie estimation.

Journal of Consumer Research, 41 (2), 261-73.

Brendl, C. M., Markman, A. B., & Messner, C. (2003). The devaluation effect: Activating a need

devalues unrelated objects. Journal of Consumer Research, 29, 463–473.

Chandon, P., & Ordabayeva, N. (2009). Supersize in one dimension, downsize in three

dimensions: Effects of spatial dimensionality on size perceptions and preferences.

Journal of Marketing Research, 46 (6), 725–738.

Chandon, P., & Wansink, B. (2006). How biased household inventory estimates distort shopping

and storage decisions. Journal of Marketing, 70 (October), 118–35.

——— (2007a). Is obesity caused by calorie underestimation? A psychophysical model of meal

size estimation. Journal of Marketing Research, 44 (February), 84–99.

——— (2007b). The biasing health halos of fast food restaurant health claims: Lower calorie

estimates and higher side-dish consumption intentions. Journal of Consumer Research,

34 (October), 301–14.

——— (2012). Does food marketing need to make us fat? A review and solutions. Nutrition

Reviews, 70, 571-593.

Chernev, A., & Gal, D. (2010). Categorization effects in value judgments: Averaging bias in

22

evaluating combinations of vices and virtues. Journal of Marketing Research, 47 (4),

738–747.

Collins, K. (1996). New Survey on Portion Size: Americans Still Cleaning Plates. Washington,

DC: American Institute for Cancer Research.

Cornil, Y., & Chandon, P. (2015a). Pleasure as a substitute for size: How multisensory imagery

can make people happier with smaller food portions. Journal of Marketing Research, in

press.

Cornil, Y., & Chandon, P. (2015b). Pleasure as an Ally of Healthy Eating? Contrasting Visceral

and Epicurean Eating Pleasure and Their Association with Portion Size Preferences and

Wellbeing. Appetite, in press.

Cornil, Y. Ordabayeva, N., Kaiser, U., Weber, B., & Chandon, P. (2014). The acuity of vice:

Attitude ambivalence improves visual sensitivity to increasing portion sizes. Journal of

Consumer Psychology, 24 (2), 177-87.

Deng, X., & Kahn, B. E. (2009). Is your product on the right side? The “location effect” on

perceived product heaviness and package evaluation. Journal of Marketing Research, 46

(6), 725-738.

Deng, X., & Srinivasan, R. (2013). When do transparent packages increase (or decrease) food

consumption. Journal of Marketing, 77 (4), 104-117.

Deutsch, C. H. (2007). Incredible shrinking packages. The New York Times, (May 12), (accessed

August 12, 2009). Available at http://www.nytimes.com/2007/05/12/business/12package.

html.

Dubois, D., Rucker, D. D., & Galinsky, A. D. (2010). The accentuation bias: money literally

looms larger (and sometimes smaller) to the powerless. Social Psychological and

23

Personality Science, 4, 199–205.

Folkes, V., & Matta, S. (2004). The effect of package shape on consumers’ judgments of product

volume: Attention as a mental contaminant. Journal of Consumer Research, 31 (2), 390-

401.

Garbinsky, E. N., Morewedge, C. K., & Shiv, B. (2014). Interference of the end: Why recency

bias in memory determines when a food is consumed again. Psychological Science, 25,

1466-1474.

Grynbaum, M. M. (2014). New York’s ban on big sodas is rejected by final court. The New York

Times, June 26.

Horovitz, B. (2007). Friday’s, Subway tailor meals for health-conscious. USA Today, March 1.

Keller, P.A., Lipkus, I.M., & Rimer, B.K. (2002). Depressive realism and health risk accuracy:

the negative consequences of positive mood. Journal of Consumer Research, 29, 57-69.

Krider, R. E., Raghubir, P, & Krishna, A. (2001). Pizzas: Pi or square? Psychophysical biases in

area comparisons. Marketing Science, 20 (4), 405–425.

Krishna, A. (2007). Spatial perception research: An integrative review of length, area, volume,

and number perception. In M. Wedel & R. Pieters (Eds.) Visual Marketing: From

Attention to Action (167-93). Mahwah, NJ: Lawrence Erlbaum Associates.

——— (2012). An integrative review of sensory marketing: Engaging the senses to affect

perception, judgment and behavior. Journal of Consumer Psychology, 22 (3), 333–51.

Ledikwe, J. H., Ello-Martin, J., & Rolls, B. J. (2005). Modifying the food environment: Energy

density, food costs, and portion size. Journal of Nutrition, 135 (4), 905–909.

Lennard, D., Mitchell, V. W., McGoldrick, P., & Betts, E. (2001). Why consumers under-use

food quantity indicators. International Review of Retail, Distribution & Consumer

24

Research, 11 (2), 177–99.

Levin, I. P., & Gaeth, G. J. (1988). How consumers are affected by the framing of attribute

information before and after consuming the product. Journal of Consumer Research, 15,

374-378.

Mohan, B., Chandon, P., & Riis, J. (2015). Percentage cost discounts always beat percentage

benefit bonuses: Helping consumers evaluate nominally equivalent percentage changes.

Journal of Marketing Behavior, 1 (1), 75-107.

Morewedge, C. K., Huh, Y. E., & Vosgerau (2010). Thought for food: Imagined consumption

reduces actual consumption. Science, 303, 1530-1533.

Nason, R. W. & Della Bitta, A. J. (1983). The incidence and consumer perceptions of quality

surcharges. Journal of Retailing, 59 (2), 40–54.

Nestle, M. (2003). Increasing portion sizes in American diets: More calories, more obesity.

Journal of the American Dietetic Association, 103, 39–40.

Nielsen, S. J., & Popkin, B. M. (2003). Patterns and trends in food portion sizes, 1977-1998.

Journal of American Medical Association, 289 (4), 450–53.

Oakes M. E. Stereotypical thinking about foods and perceived capacity to promote weight gain.

Appetite, 44(3), 317-324.

Ordabayeva, N., & Chandon, P. (2013). Predicting and managing consumers' package size

impressions. Journal of Marketing, 77, 123–137.

Papies, E. K., Stroebe, W., & Aarts, H. (2008). The allure of forbidden food: On the role of

attention in self-regulation. Journal of Experimental Social Psychology, 44, 1283–1292.

Pham, M. (2007). Emotion and rationality: A critical review and interpretation of empirical

evidence. Review of General Psychology, 11 (2), 155-178.

25

Raghubir, P. (2007). Are visual perception biases hard-wired? In M. Wedel & R. Pieters (Eds.)

Visual Marketing: From Attention to Action (143-66). Mahwah, NJ: Lawrence Erlbaum

Associates.

Raghubir, P., & Kirshna, A. (1999). Vital dimensions in volume perception: Can the eye fool the

stomach? Journal of Marketing Research, 36, 313–326.

Raghunathan, R., Naylor, R. W., & Hoyer, W. D. (2006). The unhealthy = tasty intuition and its

effects on taste inferences, enjoyment, and choice of food products. Journal of

Marketing, 70, 170–184.

Rolls B. J., Morris E. L., Roe L. S. (200). Portion size of food affects energy intake in normal-

weight and overweight men and women. American Journal of Clinical Nutrition, 76(6),

1207-121.

Rozin, P., Ashmore, M., & Markwith, M. (1996). Lay American conceptions of nutrition: Dose

insensitivity, categorical thinking, contagion, and the monotonic mind. Health

Psychology, 15 (6), 438–447

Scott, M. L., Nowlis, S. S., Mandel, N., & Morales, A. C. (2008). The effects of reduced food

size and package size on the consumption behavior of restrained and unrestrained eaters.

Journal of Consumer Research, 35 (3), 391–405.

Spence, C., & Piqueras-Fiszman, B. (2014). The Perfect Meal: The Multisensory Science of

Food and Dining. Chichester, West Sussex, UK: John Wiley & Sons Inc.

Stevens, S. S. (1986). Psychophysics: Introduction to Its Perceptual, Neural, and Social

Prospects. Oxford: Transaction Books.

Thompson, T. G., & Veneman, A. M. (2005). Dietary Guidelines for Americans (6th ed.).

Washington DC: US Government Printing Office.

26

Vanhuele, M., & Drèze, X. (2002). Measuring the price knowledge shoppers bring to the store.

Journal of Marketing, 66, 72.

Van Ittersum, K., & Wansink, B. (2012). Plate size and color suggestibility: The Delboeuf

illusion’s bias on serving and eating behavior. Journal of Consumer Research, 39, 215–

228.

van Koningsbruggen, G. M., Stroebe, W., & Aarts, H. (2011). Through the eyes of dieters:

Biased size perception of food following tempting food primes. Journal of Experimental

Social Psychology, 47, 293–299.

Viswanathan, M., Rosa, J. A., & Harris, J. E. (2005). Decision making and coping of

functionally illiterate consumers and some implications for marketing management.

Journal of Marketing, 69 (January), 15–31.

Vohs, K. D., Wang, Y., Gino, F., & Norton, M. I. (2013). Rituals enhance consumption.

Psychological Science, 24 (9), 1714-1721.

Wadhwa, M., Shiv, B., & Nowlis, S. M. (2008). A bite to whet the reward appetite: The

influence of sampling on reward-seeking behaviors. Journal of Marketing Research, 45,

403–413.

Wansink, B. (1996). Can package size accelerate usage volume? Journal of Marketing, 60 (July),

1–14.

Wansink, B., & Chandon, P. (2006a). Meal size, not body size, explains errors in estimating the

calorie content of meals. Annals of Internal Medicine, 145, 326-332.

——— (2006b). Can ‘low-fat’ nutrition labels lead to obesity? Journal of Marketing Research,

43 (November), 605–17.

——— (2014). Slim by design: Redirecting the accidental drivers of mindless overeating.

27

Journal of Consumer Psychology, 24, 413-431.

Wansink, B., & Park, S. B. (2002). Sensory suggestiveness of labeling: Do soy labels bias taste?

Journal of Sensory Studies, 17, 483-491.

Wansink, B., & Van Ittersum, K. (2003). Bottoms up! The influence of elongation on pouring

and consumption volume. Journal of Consumer Research, 30, 455–463.

Young, L. R., & Nestle, M (1998). Variation in perceptions of a ‘medium’ food portion:

Implications for dietary guidance. The Journal of American Dietetic Association, 98

(April), 458-59.

——— (2003). Expanding portion sizes in the US marketplace: Implications for nutrition

counseling. The Journal of American Dietetic Association, 103 (February), 231-34.

28

TABLE 1:

SUMMARY OF PACKAGE AND PORTION SIZE ESTIMATION BIASES AND REMEDIES

Bias Findings Remedies References Underestimation Size estimations follow

an inelastic power function of actual size, resulting in the general underestimation of package and portion sizes, more so for large sizes than for small ones.

- Information about size perception biases, attention to product, and incentives rewarding estimation accuracy do not reduce the underestimation bias. - Piecemeal estimation of individual meal components prior to estimation of total meal size reduces the underestimation bias.

Chandon & Wansink (2006; 2007a) Cornil & Chandon (2015a, 2015b) Stevens (1986) Krider, Raghubir, & Krishna (2001) Krishna (2007)

Dimensionality Size estimations are less sensitive when packages or portions change in three spatial dimensions than when they change in one dimension. This happens because people use an additive, instead of a multiplicative, model to compound changes in multiple dimensions.

- Changing packages or portions along one dimension reduces the dimensionality bias and boosts preferences for product supersizing. - Changing packages or portions along three dimensions, especially in opposite directions, increases acceptance of product downsizing. - Facilitating visual package or portion evaluation with non-visual information (e.g., weighing products by hand) reduces the dimensionality bias.

Chandon & Ordabayeva (2009) Krishna (2007) Ordabayeva & Chandon (2013)

Labeling Estimations of package and portion size are lower when labels communicate the food’s small size, healthy ingredients, healthy positioning or depict the product on the left, top, or top-left (vs. right, bottom, or bottom-right) of the package facing.

- Providing information about the food’s actual serving size reduces the effect of “low-fat” labels on estimations of normal-weight individuals, but not on estimations of overweight individuals. - Directing attention to the size, rather than the healthiness, of individual components of a combined (healthy and unhealthy) meal, improves the accuracy of size estimations for combined meals. - Encouraging consumers to consider counterfactuals (why a meal is not unique to a particular restaurant or retailer) improves the accuracy of size estimations for meals served at restaurants and retailers with healthy positioning.

Aydınoğlu & Krishna (2011) Chandon & Wansink (2007, 2007b) Chernev & Gal (2010) Deng & Khan (2009) Wansink & Chandon (2006)

29

- Positioning a package that depicts the product in the “heavy” location on the label (bottom, right, bottom-right) next to packs that depict products in the similar “heavy” locations (and vice versa) within the store shelf display attenuates the bias created by the product’s image location on the label.

Affective Size estimations are more sensitive to changes in package and portion size when consumers experience an emotional conflict towards food (i.e. when the desire to eat delicious food conflicts with perceptions that the food is unhealthy).

- Individuals who inherently feel more conflicted towards food (restrained eaters evaluating hedonic food) report more accurate estimations of package and portion size increases than individuals who do not feel conflicted towards food (unrestrained eaters evaluating hedonic food, restrained or unrestrained eaters evaluating utilitarian food). - Inducing conflict towards food (by simultaneously encouraging consumers to sample the food and highlighting that the food is unhealthy) improves the sensitivity of size estimations.

Balcetis & Dunning (2010) Cornil, Ordabayeva, Kaiser, Weber, & Chandon (2014) van Koningsbruggen, Stroebe & Aarts (2011)

30

FIGURE 1

THE UNDERESTIMATION BIAS IS REDUCED WITH PIECEMEAL ESTIMATION

(OBSERVED GEOMETRIC MEANS, 95% CONFIDENCE INTERVALS, AND MODEL

PREDICTIONS)

Note: Actual and estimated calories of small and large fast-food meals (determined via median split for regular-weight or overweight consumers). Increasing meal size, not body size (designated by BMI, body-mass index), leads to the underestimation of meal size, but separate estimation of the sandwich, side, and beverage contained in the meal eliminates the bias. Adapted from Figure 3 in Chandon, P., & Wansink B. (2007). Is obesity caused by calorie underestimation? A psychophysical model of meal size estimation. Journal of Marketing Research, 44 (1), 84-99.

0

500

1000

1500

2000

0 500 1000 1500 2000

Est

imat

ed N

umbe

r of C

alor

ies

Actual Number of Calories

Control condition,BMI < 25 (observed)

Control condition, BMI ≥ 25 (observed)

Control condition(model)

Piecemeal condition(model)

b = .83

b = .38

31

FIGURE 2

THE DIMENSIONALITY BIAS (OBSERVED GEOMETRIC MEANS, 95% CONFIDENCE

INTERVALS, AND MODEL PREDICTIONS)

Note: The underestimation of increasing sizes is stronger when objects increase in all three dimensions than when they increase in only one dimension. Adapted from Figure 2 in Chandon, P., & Ordabayeva, N. (2009). Supersize in one dimension, downsize in three dimensions: Effects of spatial dimensionality on size perceptions and preferences. Journal of Marketing Research, 46 (6), 725–738.

0

6

12

18

24

30

36

0 4 8 12 16 20 24 28 32 36

Estim

ated

Siz

e (m

ultip

le o

f the

sm

alle

st s

ize)

Actual Size (multiple of the smallest size)

1D (observed)

1D (predicted)

3D (observed)

3D (predicted)b = .93

b = .68

32

FIGURE 3

THE LABELING BIAS: HEALTH HALOS IN THE ESTIMATION OF THE SIZE OF FAST-

FOOD MEALS (OBSERVED GEOMETRIC MEANS, 95% CONFIDENCE INTERVALS,

AND MODEL PREDICTIONS)

Note: Actual and estimated calories of the four quartiles of meals from restaurants perceived to be healthy (Subway) and those perceived to be unhealthy (McDonald’s). Meals from Subway are perceived to contain fewer calories than same-size meals from McDonald’s. Adapted from Figure 3 in Chandon, P., & Wansink B. (2007). The biasing health halos of fast food restaurant health claims: Lower calorie estimates and higher side-dish consumption intentions. Journal of Consumer Research, 34 (October), 301–14.

b = .56

b = .49

33

FIGURE 4

THE LABELING BIAS IS ELIMINATED WITH COUNTERFACTUAL THINKING

Note: The perceived calorie content of the same meal is estimated to be lower when it is served in a healthy fictional restaurant, but the effect disappears after counterfactual prompting. Adapted from Figure 4 in Chandon, P., & Wansink B. (2007). The biasing health halos of fast food restaurant health claims: Lower calorie estimates and higher side-dish consumption intentions. Journal of Consumer Research, 34 (October), 301–14.

34

FIGURE 5

THE AFFECTIVE BIAS (GEOMETRIC MEANS AND MODEL PREDICTIONS)

Note: Ambivalence (the emotional conflict between the desire for food and the perceived unhealthiness of food) increases the sensitivity to increasing portion sizes. The observed sizes were obtained by computing the geometric means of the size estimations for each portion for (i) the top-quartile of restrained eaters in the regular chips condition (high ambivalence), (ii) the bottom-quartile of restrained eaters in the low-fat chips condition (indifference), (iii) the top-quartile of restrained eaters in the low-fat chips condition (low ambivalence condition) and (iv) the bottom-quartile of restrained eaters in the regular chips condition (low ambivalence condition). Adapted from Figure 3 in Cornil, Y. Ordabayeva, N., Kaiser, U., Weber, B., & Chandon, P. (2014). The acuity of vice: Attitude ambivalence improves visual sensitivity to increasing Portion Sizes.” Journal of Consumer Psychology, 24 (2), 177-87.

0

8

16

24

32

0 8 16 24 32

Estim

ated

Siz

e (m

ultip

le o

f sm

alle

st w

eigh

t)

Actual Size (multiple of smallest weight)

High ambivalence (restrained eaters and regular chips)

Indifference (unrestrained eaters and low-fat chips)

Low ambivalence (unrestrained eaters and regular chips or res-trained eaters and low-fat chips)

PredictedObserved

PredictedObserved

Predicted (both)Observed (restrained eaters and low-fat chips)Observed (unrestrained eaters and regular chips)

b = .79

b = .75

b = .70