Skip to main content
Log in

Salience and limited attention

  • Original Paper
  • Published:
Social Choice and Welfare Aims and scope Submit manuscript

Abstract

In this study, we investigate how the salience of alternatives influences the attention of a decision maker in a natural extension of the model of choice with limited attention by Masatlioglu et al. (Am Econ Rev 102(5):2183–2205, 2012). We propose positive salience effects on the attention of a decision maker. For example, when an alternative to which a consumer pays attention becomes more salient in terms of the extent of product advertisement, she will still pay attention to it in a market. We manipulate the salience of alternatives to find choice reversals from observable choice data, which play an important role in our analysis. The advantages of our model over the Masatlioglu et al. (Am Econ Rev 102(5):2183–2205, 2012) model are that it broadens the application of the model of choice with limited attention, and improves the power of inferring the decision maker’s preference and the alternatives to which she pays attention and does not pay attention.

This is a preview of subscription content, log in via an institution to check access.

Access this article

Price excludes VAT (USA)
Tax calculation will be finalised during checkout.

Instant access to the full article PDF.

Institutional subscriptions

Similar content being viewed by others

Notes

  1. This problem is called a “Type 3 failure” by Manzini and Mariotti (2014). That is, multiple preferences are compatible with the identified model with bounded rationality.

  2. Alternatively, MNO propose looking for additional data other than choice data, imposing additional restrictions on consideration sets, and using other methods, as long as the resulting revealed preferences are consistent with theirs. See also Masatlioglu and Nakajima (2015).

  3. Chetty et al. (2009) state that “without the (tax-inclusive price) tags, nearly all survey respondents ignored taxes when calculating the total price of a basket of goods, whereas with the tags, the vast majority computed the total tax-inclusive price correctly” (p. 1146).

  4. For example, a person who only considers the items on the first page of a Google search does not change her attention when an item is removed from the second page of the search. See the MNO paper for further discussions on attention filters.

  5. One possible interpretation of the choice anomaly is the “attraction effect”. (e.g., see Huber et al. 1982; Lehmann and Pan 1994). That is, z is a decoy of x, and the DM does not pay attention to x without z.

  6. See also Cherepanov et al. (2013), Lleras et al. (2017), and Manzini and Mariotti (2007, 2012) for other two-stage choice procedures.

  7. Rubinstein and Salant (2012) investigate how to elicit individual’s underlying preferences in a related framework.

  8. See Bettman et al. (1998) for a comprehensive discussion in the literature on psychological marketing and consumer research, in which a consumer’s preference plays no role in decision-making. On the other hand, Mandler et al. (2012) show a person who has a checklist is nearly equivalent to a utility maximizer.

  9. If \(x_{k}\notin F(A)\cup F(B)\), then we can obtain \(c_{ik}\) by setting \(c_{ik}=\min \{a_{ik},b_{ik}\}\) because \(x_{k}\) does not attract attention at A and B.

  10. See, for example, Häubl and Trifts (2000), Parra and Ruiz (2009), Senecal and Nantel (2004), Schoormans and Robben (1997), and Vermeulen and Seegers (2009) for experimental studies on how the manipulation of external stimuli (e.g., search tools, product recommendations, hotel reviews, and package design) influences consumers’ attention and choices in online and instore shopping environments.

  11. See Clippel and Eliaz (2012), Gerasimou (2016), Lombardi (2009), and Ok et al. (2015) for other studies that are compatible with the attraction effect.

  12. See Loomes et al. (1991), May (1954), and Tversky (1969) for experimental studies in which cyclical choice patterns are observed.

  13. This problem is applied not only to the MNO model, but also to other choice models with bounded rationality, such as Manzini and Mariotti (2007, 2012).

References

  • Adachi T (2013) Occupational gender stereotypes: Is the ratio of women to men a powerful determinant? Psychol Rep Sociocult Issues Psychol 112:640–650

    Google Scholar 

  • Bettman JR, Luce MF, Payne JW (1998) Constructive consumer choice processes. J Consum Res 25:187–217

    Article  Google Scholar 

  • Bordalo P, Gennaioli N, Shleifer A (2013) Salience and consumer choice. J Polit Econ 121:803–843

    Article  Google Scholar 

  • Cherepanov V, Feddersen T, Sandroni A (2013) Rationalization. Theor Econ 8:775–800

    Article  Google Scholar 

  • Chetty R, Looney A, Kroft K (2009) Salience and taxation: theory and evidence. Am Econ Rev 99:1145–1177

    Article  Google Scholar 

  • de Clippel G, Eliaz K (2012) Reason-based choice: a bargaining rationale for the attraction and compromise effects. Theor Econ 7:125–162

    Article  Google Scholar 

  • de Clippel G, Rozen K (2014) Bounded rationality and limited datasets. Cowles Foundation Discussion Paper 1853

  • Gaucher D, Friesen J, Kay AC (2011) Evidence that gendered wording in job advertisements exists and sustains gender inequality. J Personal Soc Psychol 101:109–128

    Article  Google Scholar 

  • Gerasimou G (2016) Partially dominant choice. Econ Theory 61:127–145

    Article  Google Scholar 

  • Häubl G, Trifts V (2000) Consumer decision making in online shopping environments: the effects of interactive decision aids. Mark Sci 19:4–21

    Article  Google Scholar 

  • Huber J, Payne JW, Puto C (1982) Adding asymmetrically dominated alternatives: violations of regularity and the similarity hypothesis. J Consum Res 9:90–98

    Article  Google Scholar 

  • Lehmann DR, Pan Y (1994) Context effects, new brand entry, and consideration sets. J Mark Res 31:364–374

    Article  Google Scholar 

  • Lleras JS, Masatlioglu Y, Nakajima D, Ozbay E (2017) When more is less: limited consideration. J Econ Theory 170:70–85

    Article  Google Scholar 

  • Lombardi M (2009) Reason-based choice correspondences. Math Soc Sci 57:58–66

    Article  Google Scholar 

  • Loomes G, Starmer C, Sugden R (1991) Observing violations of transitivity by experimental methods. Econometrica 59:425–439

    Article  Google Scholar 

  • Mandler M, Manzini P, Mariotti M (2012) A million answers to twenty questions: choosing by checklist. J Econ Theory 147:71–92

    Article  Google Scholar 

  • Manzini P, Mariotti M (2007) Sequentially rationalizable choice. Am Econ Rev 97:1824–1839

    Article  Google Scholar 

  • Manzini P, Mariotti M (2012) Categorize then choose: boundedly rational choice and welfare. J Eur Econ Assoc 10:1141–1165

    Article  Google Scholar 

  • Manzini P, Mariotti M (2014) Welfare economics and bounded rationality: the case for model-based approaches. J Econ Methodol 21:343–360

    Article  Google Scholar 

  • Masatlioglu Y, Nakajima D (2015) Completing incomplete revealed preference under limited attention. Jpn Econ Rev 66:285–299

    Article  Google Scholar 

  • Masatlioglu Y, Nakajima D, Ozbay EY (2012) Revealed attention. Am Econ Rev 102:2183–2205

  • May KO (1954) Intransitivity, utility, and the aggregation of preference patterns. Econometrica 22:1–13

    Article  Google Scholar 

  • Ok EA, Ortoleva P, Riella G (2015) Revealed (p)reference theory. Am Econ Rev 105:299–321

    Article  Google Scholar 

  • Parra JF, Ruiz S (2009) Consideration sets in online shopping environments: the effects of search tool and information load. Electron Commer Res Appl 8:252–262

    Article  Google Scholar 

  • Rubinstein A, Salant Y (2012) Eliciting welfare preferences from behavioural data sets. Rev Econ Stud 79:375–387

    Article  Google Scholar 

  • Salant Y, Rubinstein A (2008) (A, f): Choice with frames. Rev Econ Stud 75:1287–1296

    Article  Google Scholar 

  • Samuelson PA (1938) A note on the pure theory of consumer’s behaviour. Economica 5:61–71

    Article  Google Scholar 

  • Senecal S, Nantel J (2004) The influence of online product recommendations on consumers’ online choices. J Retail 80:159–169

    Article  Google Scholar 

  • Schoormans JPL, Robben HSJ (1997) The effect of new package design on product attention, categorization and evaluation. J Econ Psychol 18:271–287

    Article  Google Scholar 

  • Thaler RH, Sunstein CR (2008) Nudge: improving decisions about health, wealth, and happiness. Yale University Press, New Haven

    Google Scholar 

  • Tversky A (1969) Intransitivity of preferences. Psychol Rev 76:31–48

    Article  Google Scholar 

  • Tyson CJ (2015) Satisficing behavior with a secondary criterion. Soc Choice Welf 44:639–661

    Article  Google Scholar 

  • Vermeulen IE, Seegers D (2009) Tried and tested: the impact of online hotel reviews on consumer consideration. Tour Manag 30:123–127

    Article  Google Scholar 

Download references

Acknowledgements

This paper was circulated under the title “Generalized revealed attention”. The author is grateful to the two anonymous referees of this journal, Susumu Cato, Michele Lombardi, Marco Mariotti, Eiichi Miyagawa, Daisuke Nakajima, Koji Shirai, Christopher Tyson, Takahiro Watanabe, and Naoki Yoshihara for their valuable suggestions. This research is supported by a Grant-in-Aid for Young Scientists (B) from the Ministry of Education, Culture, Sports, Science and Technology (No. 26780125). Of course, any remaining errors are the author’s own.

Author information

Authors and Affiliations

Authors

Corresponding author

Correspondence to Yukinori Iwata.

Appendix

Appendix

We have already proved the if-parts of Theorems 1 and 2 in the body of the paper, which will be useful in proving the remaining results.

Proof of Lemma 1

We first show that if c satisfies WARP(SLA), then P is acyclic. Suppose P has a cycle: \(x^{1}Px^{2}P\cdots Px^{t}Px^{1}\). Without loss of generality, it is possible to rearrange the order of alternatives such as \(x_{1}Px_{2}P\cdots Px_{t}Px_{1}\). Then, for any \(j\in \{1,\ldots ,t-1\}\), there exist \(B^{j}\in \mathscr {A}\), \(i\in Q\), and \(b'_{i(j+1)}\in I_{i}\), with \(b^{j}_{i(j+1)}\) \(\ge b'_{i(j+1)}\), such that \(c(B^{j})=x_{j}\ne c(B^{j}\backslash b'_{i(j+1)})\), and there exist \(B^{t}\in \mathscr {A}\), \(i\in Q\), and \(b'_{i1}\in I_{i}\) with \(b^{t}_{i1}\ge b'_{i1}\) such that \(c(B^{t})=x_{t}\ne c(B^{t}\backslash b'_{i1})\). Consider a profile A with \(\{x_{1},\ldots ,x_{t}\}=F(A)\). Then, for any \(x_{k}\in F(A)\), there exists B with \(x_{k}\in F(B)\), such that \(c(B)\in F(A)\), and \(c(B)\ne c(B\backslash b'_{ik})\), for some \(i\in Q\) and some \(b'_{ik}\in I_{i}\), with \(b_{ik}\ge b'_{ik}\) but \(c(B)\ne x_{k}\), which implies that c violates WARP(SLA).

Next, we show that if P is acyclic, then c satisfies WARP(SLA). Suppose P is acyclic. Then, for any \(A\in \mathscr {A}\), there exists at least one alternative \(x_{j}\in F(A)\) such that there is no \(x_{k}\in F(A)\) with \(x_{k}Px_{j}\), which implies that there is no \(x_{k}\in F(A)\) such that \(c(B)=x_{k}\ne c(B\backslash b'_{ij})\) for some \(B\in \mathscr {A}\), some \(i\in Q\), and some \(b'_{ij}\in I_{i}\), with \(b_{ij}\ge b'_{ij}\). Therefore, for any \(B\in \mathscr {A}\) with \(x_{j}\in F(B)\), whenever \(c(B)\in F(A)\) and \(c(B)\ne c(B\backslash b'_{ij})\), for some \(i\in Q\) and some \(b'_{ij}\in I_{i}\), with \(b_{ij}\ge b'_{ij}\), we have \(c(B)=x_{j}\), which implies that c satisfies WARP(SLA). \(\square \)

Proof of Theorem 3

To show that c satisfies WARP(SLA), suppose c is a CSLA represented by \((\succ ,\Gamma )\). Then, the if-part of Theorem 1 implies that \(\succ \) includes P. Hence, P must be acyclic. It follows from Lemma 1 that c satisfies WARP(SLA).

We now suppose c satisfies WARP(SLA). Since P is acyclic by Lemma 1, we can find a preference \(\succ \) that includes P. Consider a preference \(\succ \) that includes P, and define the following consideration set mapping: for any \(A,A'\in \mathscr {A}\), \(F(A)=F(A')\) implies \(\Gamma (A)=\Gamma (A')\) and, for any \(A\in \mathscr {A}\),

$$\begin{aligned} \Gamma (A)=\{x_{j}\in F(A):c(A)\succ x_{j}\}\cup \{c(A)\}. \end{aligned}$$

Then, it is clear that c(A) is the unique \(\succ \)-best alternative in \(\Gamma (A)\). We now show that \(\Gamma \) is a salience-affected attention filter. We shall show that \(\Gamma (A)=\Gamma (A\backslash a'_{ij})\) for any \(A\in \mathscr {A}\), any \(i\in Q\), any \(x_{j}\in X\), and any \(a'_{ij}\in I_{i}\), with \(a_{ij}\ge a'_{ij}\) whenever \(x_{j}\notin \Gamma (A)\). Suppose \(x_{j}\in F(A)\), but \(x_{j}\notin \Gamma (A)\). Then, we have \(c(A)\ne x_{j}\). By construction of \(\Gamma \), we have \(x_{j}\succ c(A)\) and, therefore, \(c(A)Px_{j}\) does not hold, which implies that we have \(c(A)=c(A\backslash a'_{ij})\). By construction of \(\Gamma \), we have \(\Gamma (A)=\Gamma (A\backslash a'_{ij})\). \(\square \)

Proof of Theorem 1

We will prove the only-if part of Theorem 1. That is, we show that if \(x_{j}\) is revealed to be preferred to \(x_{k}\), then we have \(x_{j}P_{R}x_{k}\). Suppose \(x_{j}P_{R}x_{k}\) does not hold. Then, there exists a preference \(\succ \) that includes \(P_{R}\) and ranks \(x_{k}\) higher than \(x_{j}\). The proof of Theorem 3 shows that c can be represented by the preference \(\succ \). Therefore, we conclude that \(x_{j}\) is not revealed to be preferred to \(x_{k}\). \(\square \)

Proof of Theorem 2

We will prove the only-if part of Theorem 2.

(Revealed Inattention): We first show that if \(x_{j}\) is revealed not to attract attention at A, then we have \(x_{j}P_{R}c(A)\). Suppose \(x_{j}P_{R}c(A)\) does not hold. Then, consider a preference \(\succ \) that includes \(P_{R}\) and ranks c(A) higher than \(x_{j}\). By the proof of Theorem 3, c can be represented by the preference and a salience-affected attention filter, with \(x_{j}\in \Gamma (A)\).

(Revealed Attention): Next, we show that if \(x_{j}\) is revealed to attract attention at A, there exists \(B\in \mathscr {A}\) that satisfies the condition of statement 2 in Theorem 2. Suppose there exists no B that satisfies the condition. We prove that if c is a CSLA, then it can be represented by some salience-affected attention filter \(\Gamma \), with \(x_{j}\notin \Gamma (A)\). If \(c(A)P_{R}x_{j}\) does not hold, we have already shown that c can be represented by \((\succ ,\Gamma )\) with \(x_{j}\succ c(A)\) and \(x_{j}\notin \Gamma (A)\). Thus, \(x_{j}\) is not revealed to attract attention at A and, therefore, we only consider the case where \(c(A)P_{R}x_{j}\) holds.

Now construct a binary relation \(\tilde{P}\) as follows: \(a\tilde{P}b\) if and only if “\(aP_{R}b\)” or “\(c(A)=a\) and not \(bP_{R}c(A)\)”. That is, \(\tilde{P}\) ranks c(A) as high as possible unless it contradicts \(P_{R}\). Since \(P_{R}\) is acyclic and c is represented by a salience-affected attention filter, we can show that \(\tilde{P}\) is also acyclic. Given this observation, consider a preference \(\succ \) that includes \(\tilde{P}\) as well as \(P_{R}\). We have already shown that the following consideration set mapping is a salience-affected attention filter: for any \(A',A''\in \mathscr {A}\), \(F(A')=F(A'')\) implies \(\tilde{\Gamma }(A')=\tilde{\Gamma }(A'')\) and, for any \(A'\in \mathscr {A}\), \(\tilde{\Gamma }(A')\equiv \{x_{j}\) \(\in F(A'):c(A')\succ x_{j}\}\cup \{c(A')\}\). In addition, we can show that c is represented by \((\succ ,\tilde{\Gamma })\). We now define \(\Gamma \) as follows:

$$\begin{aligned} \Gamma (A')=\left\{ \begin{array}{ll} \tilde{\Gamma }(A') &{}\quad \text {for }A'\notin \Omega \\ \tilde{\Gamma }(A')\backslash x_{j} &{}\quad \text {for }A'\in \Omega \end{array}\right. , \end{aligned}$$

where \(\Omega \) is a collection of profiles, such that

$$\begin{aligned} \Omega = \begin{Bmatrix}&(\mathrm{a}) ~c(A')=c(A), \\&(\mathrm{b}) ~ x_{k}P_{R}c(A) ~ \text {for any }x_{k}\in (F(A)\backslash F(A'))\cup (F(A')\backslash F(A)), \\ A'\in \mathscr {A}:&\text {and (c) for any} ~i\in Q\backslash \{1\} \text {and any} x_{k}\in F(A)\cup F(A'), \\&a_{ik}=a'_{ik} \text {unless }x_{k}P_{R}c(A) \end{Bmatrix}. \end{aligned}$$

Thus, \(\Gamma \) is obtained from \(\tilde{\Gamma }\) by removing \(x_{j}\) from \(\tilde{\Gamma }(A')\) of any profile \(A'\in \mathscr {A}\) where (a) \(c(A')=c(A)\), (b) any alternative that belongs to F(A) and \(F(A')\), but not to both, is revealed to be preferred to c(A), and (c) for any \(i\in Q\backslash \{1\}\) and any \(x_{k}\in F(A)\cup F(A')\), \(a_{ik}=a'_{ik}\), unless \(x_{k}\) is revealed to be preferred to c(A). Note that it follows from \(c(A)P_{R}x_{j}\) that \(c(A)\ne x_{j}\). Therefore, \(\Gamma (A')\subset \tilde{\Gamma }(A')\) always contains \(c(A')\). Furthermore, since \((\succ ,\tilde{\Gamma })\) represents c, \((\succ ,\Gamma )\) also represents c. Thus, we only need to show that \(\Gamma \) is a salience-affected attention filter.

Note that \(\tilde{\Gamma }\) is a salience-affected attention filter and \(c(A')=c(A'')\) whenever \(\tilde{\Gamma }(A')=\tilde{\Gamma }(A'')\) because \((\succ ,\tilde{\Gamma })\) represents c.

Suppose \(x_{k}\notin \Gamma (B)\) for any \(B\in \mathscr {A}\). We prove \(\Gamma (B)=\Gamma (B\backslash b'_{ik})\) for any \(i\in Q\) and any \(b'_{ik}\in I_{i}\), with \(b_{ik}\ge b'_{ik}\). We now show that \(\Gamma (B)=\Gamma (B\backslash b'_{ik})\), for any \(i\in Q\) and any \(b'_{ik}\in I_{i}\) with \(b_{ik}\ge b'_{ik}\). We distinguish three possible cases: (i) \(x_{k}=x_{j}\), (ii) \(B\in \Omega \) and \(x_{k}\ne x_{j}\), and (iii) \(B\notin \Omega \) and \(x_{k}\ne x_{j}\).

Case (i): If \(B\notin \Omega \), then we have \(\Gamma (B)=\tilde{\Gamma }(B)=\tilde{\Gamma }(B\backslash b'_{ik})=\Gamma (B\backslash b'_{ik})\). If \(B\in \Omega \), then \(c(B)=c(B\backslash b'_{ik})\) must hold (otherwise, the condition of the statement is satisfied). It follows from the construction of \(\tilde{\Gamma }\) and \(\Gamma \) that \(\Gamma (B)=\tilde{\Gamma }(B)\backslash x_{j}=\tilde{\Gamma }(B\backslash b'_{ik})=\Gamma (B\backslash b'_{ik})\).

Case (ii): Since \(x_{k}\notin \Gamma (B)\) is equivalent to \(x_{k}\notin \tilde{\Gamma }(B)\), we have \(\tilde{\Gamma }(B)=\tilde{\Gamma }(B\backslash b'_{ik})\), by construction of \(\tilde{\Gamma }\). Therefore, we have \(c(B\backslash b'_{ik})=c(B)=c(A)\). Since we have \(x_{k}\notin \Gamma (B)\), \(x_{k}\succ c(A)\) must hold, by construction of \(\Gamma \) and \(\tilde{\Gamma }\), which implies \(x_{k}P_{R}c(A)\) by construction of \(\succ \). Then, we obtain \(B\backslash b'_{ik}\in \Omega \). Therefore, we have \(\Gamma (B)=\tilde{\Gamma }(B)\backslash x_{j}=\tilde{\Gamma }(B\backslash b'_{ik})\backslash x_{j}=\Gamma (B\backslash b'_{ik})\).

Case (iii): If \(B\backslash b'_{ik}\in \Omega \), we have \(c(B)=c(B\backslash b'_{ik})=c(A)\) and \(x_{k}P_{R}c(A)\) by a similar argument to that above. Therefore, \(B\in \Omega \) must hold, which is a contradiction. Thus, we have \(B\backslash b'_{ik}\notin \Omega \), which implies that \(\Gamma (B)=\tilde{\Gamma }(B)=\tilde{\Gamma }(B\backslash b'_{ik})=\Gamma (B\backslash b'_{ik})\). \(\square \)

Rights and permissions

Reprints and permissions

About this article

Check for updates. Verify currency and authenticity via CrossMark

Cite this article

Iwata, Y. Salience and limited attention. Soc Choice Welf 50, 123–146 (2018). https://doi.org/10.1007/s00355-017-1077-1

Download citation

  • Received:

  • Accepted:

  • Published:

  • Issue Date:

  • DOI: https://doi.org/10.1007/s00355-017-1077-1

Navigation