Are preferences inherent or constructed? A response to Simonson (2008)
The question of whether, or at least when and how, preferences are constructed or inherent is a hotly debated issue in decision-making.
For example, Tversky, Slovic and Kahneman (1990) show the existence of preference reversal, implying that preferences can be constructed by the context. TSK (1990) show that preference reversal can be attributed to violations of invariance, in particular, alternative elicitation procedures, in that different preferences are revealed depending on whether it is by pricing or choice.
Given preference reversal, and other empirical evidence, Bettman et al. (1998) posit that consumer choice is inherently constructive[1], whereby consumers do not have well-defined existing preferences. They proceed to provide an integrative framework, considering four relevant consumer choice goals — optimizing the accuracy-effort tradeoff, minimizing negative emotion, and maximizing ease of justification.
In response, Simonson (2008) questions the over-reaching conclusion, regarding decision making in general, of the findings that preferences are constructed, that “preferences may not exist” (p. 160). Simonson’s aim is to (re)introduce the notion and importance of stable, inherent (vs. constructive) preferences that are not determined by context.
Simonson proposes a bounded rationality account, in that preference construction may instead reflect people’s difficulty in evaluating (the trade-offs of) absolute attribute values, such that they tend to consider relative evaluations among the choice set in the attribute space (as manipulated in lab experiments). For example, a representative study that concludes that preferences may be constructed is Drolet et al. (2000), which had one group given the relative position of each option (lower or higher, fat or taste), without the specific (absolute) attribute options. (Another issue is the potential leading nature of the measure, which asks subjects to give a likelihood to select the compromise option, instead of a choice, pricing, or rating).
However, as Simonson says, and I am inclined to agree, this does not mean that absolute values or locations in the attribute space do not matter at all. Following from the Drolet et al. (2000) study, a preference for “high fat” ice cream — because it has better taste — has a plausible upper (absolute) limit, at least for the attribute of fat content. I would think that most of us would not be willing to eat a small cup of ice cream with, say, 1 tonne of fat content, no matter how tasty compared to alternatives. Perhaps we can conceive of this as being the boundary conditions of preferences “travel[ling] with the choice set”, within which (fat content ≤ 1 tonne) preferences can be constructed by context, but outside of which, there are no possible preference that can exist, and thus, preferences writ-large, are inherent in that sense, bound within some set in the universal space of all possible choices.
In the same breath (“in particular”), Simonson critiques the generalizability (which he equates, or argues it follows from, external validity) of such findings from lab experiments to the psychology of decision-making and real-world consumer preferences in general. For example, he argues that the evidence of preference reversal based on choosing vs. pricing gambles, as demonstrated by TSK (1990), are “unusual tasks”, and given the specific characteristics and nature of the pricing task itself, lead subjects to “rely on the most accessible relative judgments, namely, assign a dollar price relative to the stated dollar amounts”. As such, these experiments “capitalized on the very principles of construction they were demonstrating”, specifically, the “manipulation of the salient reference points” (p. 158).
This, however, I propose humbly, is a distinct, and weaker, critique. The former distinguishes between absolute and relative valuation — which is undoubtedly true — and thus provides a good distinction between inherent and constructed preferences. The latter is a more abstract, conceptual critique of whether preferences can be understood with or without its measurement or elicitation method (which I consider later).
Simonson is right when he says that the reversal due to different elicitation methods is an unhelpful definition of constructed preferences, at least if only (mainly) one is used in real life, but the converse, that preferences can be stable in spite of different manifestations due to different elicitation methods, means that those preferences are unidentifiable and so, is unhelpful too. In theory, we could have any and all stable preferences in us all along, but which we would never know without measurement of some kind.
How then might we investigate the inherent or constructed nature of preferences? One potential example is Tu & Hsee (2016). Similarly, they distinguish betwen two types of preferences, inherent (stable) and learned (acquired and variable), but unlike Simonson (2008), inherence, as defined here, is not the result of individual genetic differences, but evolutionarily-formed and thus stable (‘hard-wired’ in mind and body) whereas constructed (“learned”) preferences are acquired more recently and rely on social comparison, and thus “malleable across time and contexts”.
To identify whether inherence or construction (in the form of learnedness) is at play, the authors cleverly test the same preference object, but showing how the nature varies depending on the resource state of the subject. This may be one method to show when and how preferences are inherent or constructed. (Though I suppose if one takes a meta-view, if preferences can be both inherent and constructed, then they must be meta-constructed.)
Here, I would like to reflect on the debate between stable and inherent preferences. In particular, I reference and largely follow the structure of Simonson (2008), since he frames his article specifically on the distinction. Other than understanding the nature and types of decision-making, whether and which preferences are inherent vs. constructed have important managerial implications, such as whether to manipulate the context, frame, or task, versus selecting consumers with preferences relevant to the firm’s offerings.
It is important to first note that Simonson does not argue that there are only stable preferences, and repeatedly cautions against this false takeaway from his article. Indeed, he quotes a few of his own papers which are based on, or support, the premise that preferences are to some extent constructed, for example, famously, Simonson & Tversky (1992). At the same time, perhaps to emphasize the importance of his proposition, that along the spectrum from stable to constructed preferences, decision-making research, at least around the time of his paper’s publication, has tended to focus too much on the latter, he also (has to) make(s) his point somewhat forcefully.
Ultimately, the inherency or constructability of preferences is, as is the case with most concepts, not categorical, but almost certainly lies on a spectrum, and the answer to which is more dominant is, as always, “it depends” — whether it’s the context, the individual etc.
The more important point, as I hope to explore later, and as Simonson hopes and calls for, is how can that distinction be meaningful.
Stability and inherent preferences
Simonson defines inherent preferences as “relatively stable preference components or dispositions that are not determined by the context, task, or frame” (p. 162).
Hence, Simonson seems to suggest that inherence implies stability. However, the reverse may not be true, which is important if we use stability as a measure of inherency. Stable preferences could be constructed given some context, but if the context (or the parts of which that is considered by the decision-maker, actual or perceived) remains relatively stable, the preferences, by definition (or construction), do not change when measured.
It may seem semantic, displacing the focus from “stable” to the context that results in it, but this is important, because construction would be shown if the context were changed, or when the context, task, or frame, affects that which is considered by the decision-maker.
Of course, over time, with learning and habituation, construction would likely be less explicit and less effortful, to the point that it may become, for all intents and purposes, inherent, and the issue of whether to consider that preference inherent vs. constructed becomes a subjective matter of where one should draw the line (if at all possible or desirable) when distinguishing or defining. Simonson also seems to suggest this: “a person who over time becomes more health conscious may correspondingly develop inherent preferences” (p. 163, emphasis added). The potential conceptual contradiction between “develop” (suggesting some form of change) and “inherent” is a testament to blurred lines between these two concepts.
This alternative explanation (unchanging preferences can also be constructed, and thus, are not necessarily inherent) becomes more apparent in the Simonson’s proposed but contradictory measures. For example, he observes “there are many things that most people do not adapt to and continue to dislike” (p. 162) when he introduces the notion of inherent preferences, suggesting that if one’s (negative) preference is stable, it is inherent. However, he then suggests that those “who adapt to [a new, or existing but unfamiliar] object [presumably to overcome unfamiliarity] and come to like it [suggesting initial negative preference] can be assumed to have had an inherent preference”. Indeed, he himself states that “Adaptation to (i.e., coming to like) attribute values might be seen as an indicator that inherent preferences either do not exist or are malleable, or alternatively, as an indicator of uncovered inherent preferences” (p. 166), though he supports the latter given “adaptation is far from universal”.
Still, it seems that one can come up with plausible accounts of how inherent preferences can lead to both unchanging and adaptative preferences, at least given an initial negative preference, so this definition of inherent preferences at least may not be measurable (and its related hypotheses thus unfalsifiable).
Indeed, all possible initial and subsequent preference (whether it changes or not, whatever its initial valence) can be justified as being inherent preferences:
Dormancy, experience, and inherent preferences
To make his case for inherent preferences, Simonson uses a personal experience, his pillow anecdote[2], as well as the introduction of new products (iPhone, Nintendo Wii) to show that some consumers may have “a dormant [used concurrently or alternatively with “inherent”, “dispositional”] preference”, which makes them “receptive” (or not) to those new products.
Thus, at least in the context of preferences towards new or unknown (to the focal consumer) offerings, Simonson seems to equate (or at least mention together) the dormancy (vs. “active and retrievable”, p. 162/3) of preferences and those preferences being inherent (vs. constructed).
In other words, if there are dormant preferences, they are inherent.
The question is whether this equation (or conflation) is necessary, and if so, whether the idea of inherent preferences may only be (mainly) meaningful in the context of new or unfamiliar products, such that there is a “half-life” to the weight of the inherent component of preference. Accounting for potential novelty and familiarization effects, as products are uncovered and become mainstays, does the inherent component become less decisive, even survive?
Simonson argues no: “Once uncovered [by experience], (previously “dormant”) inherent preferences become “active” and retrievable from memory… once active and stored in memory, they are likely to be more resistant to change than preferences acquired through construction” (p. 162).
Indeed, Simonson links, though not exclusively (please refer to the next section), inherent preferences to experience (vs. choice or pricing, perhaps due to the need of some observable measure of an otherwise untestable concept): “actual [vs. perceived] experience [by focusing attention on the stimulus] provides absolute valuations and dormant inherent preferences the best chance to emerge” (p. 163, emphasis added). He later explains that “decisions promote and highlight forces of preference construction and reference points whereas experiences focus attention on the stimulus and thus tend to put relatively more emphasis on inherent preferences” (p. 163–4, emphasis added). The importance of experiences for Simonson is further implied when he considers individual difference moderators, all of which he states are mediated by the willingness to experiment, and when he repeatedly mentions experience again in the Discussion section in the context of inherent vs. constructed preferences[3].
Importantly, coupled with his previous definition that inherent preferences are “relatively stable preference components or dispositions that are not determined by the context, task, or frame” (p. 162), Simonson thus seems to suggest that experience, and thus preference, of a pillow or iPhone, could be independent of context, at least theoretically (in his words, “I must have had in me”).
However, there is evidence that the context affects the experience itself. For example, in a clever three-group experimental design, Lee et al. (2006) found that only when the disclosure of negative information preceded (but not when it proceeded) tasting of a beer, was preference significantly reduced, which implies that disclosure affected the experience itself and thus preferences, not simply people’s post hoc assessment of the experience.
Again, while there is likely a stable component to experiences (e.g. the anatomy and physiology of taste buds), it is highly likely that experience is not merely the result of biological senses internal to us, but that our senses somehow (perhaps neurobiologically) also interact with the broader environment. Indeed, Lee and colleagues (2006) distinguish between, but not dismiss either, “bottom-up processes which reflect characteristics of the stimulus impinging on the perceiver’s sensory organs, and top-down processes, which reflect the perceiver’s beliefs, desires, and expectations”, both of which “jointly determine” the quality of an experience (p. 1054).
Non-experiential inherent preferences
Simonson does offer non-experiential contexts where inherent preferences may exist, such as political (e.g. type of healthcare system) or ethical preferences (e.g. organ donation), which he argues is less context-sensitive and based upon core values, or “matters of principle” (p. 164).
However, one interpretation is that this merely displaces (and does not answer) the debate of inherent vs. constructed to an antecedent of preference — the values informing them — where values can be seen as a type of higher-order preference (see last section).
More importantly, are such preferences really independent of context? Other than the fact that the “preference ingredients” of political preferences are complex (Simonson cites concepts of rights — “healthcare for all”, the role of government, taxation etc.) which would likely need to be learnt — thus open to construction — to the point where one can make some preference call, context itself might matter, in principle, for these preferences.
As an anecdote, in a hilarious Saturday Night Live skit about a fictional game show, “Republican or Not?”, where ‘contestants’ have to guess the political affiliation of a person espousing certain values, even in the current increasingly polarized landscape (which likely result in more stable partisan preferences), it becomes clear that a value or innate belief alone is indecisive. For example, one of the ‘clues’ provided by the person is “My body, my choice” (i.e. the value of bodily autonomy). One of the ‘contestants’, initially eager to answer, then hesitates and asks, “You’re talking about vaccines, right?” The other (unstated) preference object is abortion. (The stereotypical Republican, post-Covid, is against vaccines and pro-life.)
Cognitive dissonance aside, this suggests that even if there exists “active inherent preferences for these ingredients” (p. 164) it may not lead to (and predict) the same specific preference. A pro-abortion preference has been justified from bodily autonomy as well.
The fact that follow-up questions were asked, in the skit, to clarify these ‘preference(s) (ingredients)’ suggest that political and ethical preference ingredients, unless so specific as to render them ungeneralizable to the principles of decision-making (as desired by Simonson), must be defined in, and by, some context, and are thus not innate or inherent, at least in resulting in measurable predictions.
Indeed, the recent shift against vaccines among Republicans is evidence that they are likely, or at least could be (and did become), constructed, first beginning with those for Covid-19, before the shift was also noted in general attitudes towards vaccination (Fridman et al., 2021; Vox, 2022; Perry, 2023). More generally, empirical research has shown that political preferences can change, such as through choice blindness (Pärnamets & Van Bavel, 2018) vs. informational inputs (echo chambers etc.)
In the case of the preference change towards vaccines, this was arguably due, in part, to Trump’s idiosyncratic stance towards vaccines (hence, the context of one’s social group or leader is sine non qua for one’s own political preferences, especially as politics — being a form of organization of people — in the plural — is inherently social). In short, political preferences are other-oriented preferences, and thus constructable (by varying the ‘other’).
(I suppose one could define the preference to conform with socio-cultural norms or preference to obey authority as stable, but again, it’s so broad as to be unpredictive of the preferences towards specific preference objects like vaccines. Also, that in itself does not explain which group and which authority would be determinative. Many Republicans do not support the current Democratic administration(‘s policies).)
One may argue it’s political affiliation (Democrat or Republican) that matters here, another variable that overpowers and thus ultimately determines specific preference (people want to follow their party’s stance), but the idea is that, at least following Simonson’s process, the same ingredient may not lead to the same (inherent) political preferences, but a post-hoc construction (Haidt, 2001). (I apply moral decision-making processes here since political preference likely has a normative component.)
As such, people may pick and choose values, so broad and general as to be uncontestable, to apply to an arbitrary political preference, and retroactively use that to update their ingredient component which becomes (and they believe is) ‘innate’. It is likely most people generally value autonomy, freedom, equality, fairness etc. The differences and disputes lie in what extent and in what form. This, however, renders any “core value” (ingredient) unpredictive.
On the other hand, Haidt (2001)’s conceptualization also suggests that moral judgment is the result of quick, automatic evaluations (intuitions), which one may draw a parallel with ‘dormant (inherent) preferences’. At the same time, Haidt (2001) also proposes that moral judgment is more the result of social and cultural influences than that of reasoning, suggesting constructed preferences. (Haidt was not arguing along the lines of inherent (vs. constructed); instead, he was considering reasoning vs. intuition.)
Ultimately, the jury is still out; this topic area, significant in its own right — definitely needs more investigation (or at least clarifying and using consistent definitions).
The external validity of inherent preferences as defined by Simonson
Finally, and importantly, Simonson distinguishes “carefully designed tasks and contexts” with “reference points typically available in the marketplace” (p. 161, emphasis added), such that the former allows for the conclusion that preferences are constructed while the latter would allow for inherent preferences to be expressed.
Indeed, he argues that lab experiments result in a paucity of information (other sources, other alternatives), which inhibit making absolute judgments[4]. Thus, Simonson (2008) seems to suggest that preferences can be, or would be classified as inherent if consumers have available the full set of possible information (sources, options).
(Indeed, Fridman and colleagues [2021] suggest that one of the reasons behind Republicans’ changing political preferences regarding vaccines is because of differential exposure to media channels and social networks.)
This is technically consistent with his definition of inherent preferences being “relatively stable preference components or dispositions that are not determined by the context, task, or frame”. Since the full set of possible information is there, there is “context, task, or frame” to speak of, other than one — that of all contexts, tasks, and frames.
This itself, however, is untenable in reality. The limits and variation of memory retrieval, sensing and cognitive abilities, as well as the trust towards informational sources notwithstanding, new information is always being created and hence updated. For example, in the context of Drolet et al. (2000), new health insights on the effects of fat, options may come in (new ice cream flavors) or out (sold out discontinued).
If inherent preferences (are defined to) exist in the context of complete information, it seems they may be limited to each specific point of time, thus limiting its external validity, and importantly, its predictability. This does not even account for the varying state of the decision-maker (as Simonson also recognizes, there are “transient states of mind (e.g. mood) [and] competing judgment criteria”, p. 161). In that sense, preferences are relative across time, even if not in (attribute) space.
(One could perhaps account for time, taking into account future developments, but unless one’s view of the universe is deterministic and one could somehow account for all of it, it seems there are various plausible future sets of information, and no guarantee that all of them would result in the same, inherent preference. Given the complexity bordering on the philosophic, I leave this for now, in search of headache medication.)
Some potentially useful distinctions to understand inherent versus constructed preferences.
As Simonson advocates, one of the key purposes of revisiting stable, inherent preferences is that the distinction between it and constructed preferences should be meaningful and provide useful insights.
Some distinctions mentioned, or alluded to, by Simonson is local (constructed) vs. global (inherent), which parallels the attribute space conceptualization, where “within the boundary” preferences are constructed and “within-without the boundary” preferences are inherent (cf. sample space vs. event space in probability theory).
Another is provided by Tu and Hsee (2016) in that inherent preferences are formed evolutionarily and thus stable (‘hard-wired’ in mind and body) whereas constructed (“learned”) preferences are acquired more recently and rely on social comparison, and thus “malleable across time and contexts”.
I propose two other (not necessarily mutually exclusive) distinctions which may help. (Disclaimer: They could just be rephrasing of the previous typologies and definitions, but at least they might help clarify from a different perspective.)
First-order versus higher-order preferences
This distinction comes from philosophical idea of higher-order volition (desires), to put it in this context, there are preferences about preferences, vs. preferences themselves.
Because we are thinking creatures with willpower and the ability to engage in metacognition, we can overcome, or at least regulate, our relative thinking-prone, occurrence-specific, first-order preferences by ensuring they are consistent with an overarching, higher-order preference that we want to be stable (sigh, would this be a third-order preference? I’ll leave this for now; it can go on infinitely.)
For example, my instinctual, first-order preference is to consume tasty — but unhealthy, even if I don’t specifically desire unhealthiness — fast food, which is easily variable given the circumstances, but if I have a higher-order, more stable preference to maintain a healthy diet, I may self-regulate my first-order preferences (cravings, temptations, etc.) and thus express the higher-order preference.
This does not mean the first-order preference never materializes. It could be the case that the higher-order preference of maintaining a healthy diet allows for consumption of tasty and unhealthy fast food (a “cheat day”), for say, 20% of the time, during which time specifically, the first- and higher-order preferences are congruent.
Different order preferences would not be the same as Simonson’s ingredients (componential) conceptualization of preferences, where ingredients (e.g. taxation, government intervention, rights preference) represent a preference (e.g. healthcare system). The former is where one takes precedence over the other; the latter is that the ingredients — varying in their congruence to one another — collectively form the aggregate preference.
There’s another can of worms regarding the nature of higher-order preferences — are they inherent or constructed (cf. pushing the debate from preferences to values), which I’ll leave aside for now.
Based on (stable) capacities or compared to other options
The second distinction is whether the preference is constrained by (stable) genetic or physiological capacities vs. the external, variable context — where the capacity is stable and independent of context (disregarding, for now, Lee et al. [2006]’s and similar findings) — (“inherent”); or defined (made) in relation to other options (“constructed” by context).
As such, preferences that are dependent on, or result from, such stable capacities or tendencies (e.g. temperature, at least its range; here, other temperatures are not different options but measures) are likely stable, and by this definition, inherent.
If options (alternatives) are sine qua non to the preference under consideration, then it’s constructed. In other words, if the preference is a decision or choice of one option versus others, those other option(s) necessarily provide(s) a counterfactual to, and opportunity cost, of the focal option that is preferred to whatever extent. Hence, that preference is ipso facto relative (to the other options), thus constructed (in terms of the other options). cf. the debate, in the philosophy literature, of whether (pairwise) preferences, or desires, are basic (Schroeder, 2020).
(Technically, there is one potential counterfactual for inherent preference as defined, namely the status quo of not possessing the preference object, but this presumes the preference must entail purchase or ownership of some kind, cf. Lamberton & Goldsmith, 2020.)
I recognize however that this distinction could be controversial, since the same preference object (e.g. temperature) can be classified as both. Hence, this may just be another expression of the absolute (range) vs. relative (discrete) distinction, or the within-boundary, within-without boundary distinctions made before. However, one can compare it to resource-based preference, a same preference object whose relevant type is dependent on the decision-maker’s state (Tu & Hsee, 2016).
(There is also, I think, a subtle but distinct difference with the relative weight conceptualization. I don’t think it’s the same here, in that it is not about which dominates the other, but its typology a priori.)
In other words, one criterion to distinguish between them, at least in practice, is to whether the preference exists by itself, or (is elicited) with options or without them.
Using Simonson (2008)’s anecdote, if his preference was to what extent does he prefer the ‘medium’ pillow, period, then that is inherent. An equivalent inherent preference is to what extent he prefers a pillow made of spikes, period. No changeable context is present or considered. His preference in this elicitation is inherent. One can think of the relevant measure being the strength and valence of the preference.
However, if it’s to what extent a ‘medium’ pillow is preferred by Simonson vis-à-vis a soft or hard pillow, then that preference is constructed. An equivalent constructed preference is his soft pillow preference vis-à-vis a softer or softest pillow. The different option sets would construct, potentially differently, the preference.
Returning to the point of Simonson’s external-validity-of-elicitation-methods critique, perhaps preferences can only be understood including their elicitation methods. In other words, elicitation is sine qua non to the preference, and there would be no point — at least unidentifiable and unmeasurable — if we conceptualize preferences without an accompanying elicitation or measure. In that meta-sense, there is no preference reversal based on elicitation, since they are distinct preferences given the proposed inseparability of the content of the preference from its elicitation.
Finally, note that this distinction is theoretical. Following Lee et al. (2006)’s premise, if similar other studies, especially neuroscientific ones, show that empirically, there is no such thing, at least with meaningful distinct predictions, as stable, within-person, biological/physiological capacities, then this distinction has limited meaningful implications.
(There is evidence that one’s internal state is not wholly stable, such that there are state-dependent preferences, where preferences change due to varying levels of physiological indicators, such as hunger, e.g. Read and van Leeuwen [1998]. But the extent to which it affects preference may be moderated by one’s inherent or baseline self-control, or self-efficacy, which may be, depending on the person, to some extent stable.)
Woof, I need some headache medication. But this is just too interesting.
References
Bettman, J. R., Luce, M. F., & Payne, J. W. (1998). Constructive consumer choice processes. Journal of Consumer Research, 25(3), 187–217.
Fridman, A., Gershon, R., & Gneezy, A. (2021). COVID-19 and vaccine hesitancy: A longitudinal study. PloS one, 16(4), e0250123.
Haidt, J. (2001). The emotional dog and its rational tail: A social intuitionist approach to moral judgment. Psychological Review, 108(4), 814–834.
Lamberton, C., & Goldsmith, K. (2020). Ownership: A perennial prize or a fading goal? A curation, framework, and agenda for future research. Journal of Consumer Research, 47(2), 301–309.
Lee, L., Frederick, S., & Ariely, D. (2006). Try it, you’ll like it: The influence of expectation, consumption, and revelation on preferences for beer. Psychological Science, 17(12), 1054–1058.
Pärnamets, P., & Van Bavel, J. (2018). How political opinions change. Scientific American. Retrieved from https://www.scientificamerican.com/article/how-political-opinions-change/
Perry (2023). The Real Reason Behind Conservatives’ Shifting Views on Childhood Vaccines. TIME. Retrieved from https://time.com/6280666/conservatives-shifting-views-childhood-vaccines/
Read, D., & Van Leeuwen, B. (1998). Predicting hunger: The effects of appetite and delay on choice. Organizational Behavior and Human Decision Processes, 76(2), 189–205.
Schroeder, T. (2020). Desire. The Stanford Encyclopedia of Philosophy.
Simonson, I. (2008). Will I like a “medium” pillow? Another look at constructed and inherent preferences. Journal of Consumer Psychology, 18(3), 155–169.
Tu, Y., & Hsee, C. (2016). “Consumer Happiness Derived from Inherent Preferences versus Learned Preferences,” Current Opinion in Psychology, 10, 83–88.
Tversky, A., Slovic, P., & Kahneman, D. (1990). “The Causes of Preference Reversal,” Te American Economic Review, 80(1), 204–17.
Vox (2022). How American conservatives turned against the vaccine. Retrieved from https://www.youtube.com/watch?v=sv0dQfRRrEQ
Footnotes
[1] interestingly, as it turns out, this particular two-word phrase is an oxymoron, given the distinction in terminology made later (“inherent” versus “constructed”)
[2] Perhaps for simplicity, Simonson did not mention the potential influence of other effects, such as the endowment effect of owning the soft pillow, although perhaps he believes he is impartial enough to control for that, and the stakes are so low that he would be able to make an objective evaluation, thus the hook of his paper’s title, “Will I like a ‘medium’ pillow?”
[3] “impact of preferences formed during experience”; “if we focus on more enduring preferences that emerge after experience” and “most notably, during experience” (p. 165); “inherent preferences are most likely to be revealed based on (sufficiently long) experience” (p. 167)
[4] In Simonson’s words, “(a) consumers typically have additional information sources that facilitate making absolute judgments, and (b) they encounter more than three options, which provides more information about the key dimensions and product types, thereby decreasing the dependence on the configuration of any particular option subset.” (p. 160)
