You are on page 1of 7

A Classification of Product Comparison Agents

Yun Wan Satya Menon Arkalgud Ramaprasad


College of Business Administration
University of Illinois at Chicago
1-312-413-8090 1-312-413-4894 1-312-996-9260
ywan1@uic.edu menon@uic.edu prasad@uic.edu

Amazon’s Z-shops offered by third- party vendors who utilize the


ABSTRACT platform provided by Amazon to sell to its customer base.
Over the last decade of rise and growth of e-commerce, a large
For consumers, the task of selecting the most preferred product
number and variety of electronic comparison-shopping agents
from available alternatives is relatively easy in traditional
have emerged on the Web, but little research has been done on
environments like Wal-Mart stores, because they are choosing
classifying them. In this paper, we propose a conceptual schema
from limited choice sets. However, this task could become a
for categorizing comparison-shopping agents into three groups:
major problem in the online environment where the number of
differentiation agents, evaluation agents and preference agents.
choice alternatives could easily grow beyond the cognitive
We describe and analyze each category in detail. We suggest that
capacity of consumers [29]. Recent research studies have shown
comparison-shopping agents can be regarded as a species in the
that choice overload can easily occur in online and traditional
ecosystem of Internet. They compete and cooperate with each
shopping environments as the number of alternatives become
other for information extraction and aggregation. By extending
large, necessitating more cognitive effort in the task of
the classification, we suggest that a new species of comparison-
comparison-shopping [5, 14, 25]. We can expect that in the near
shopping agent will emerge driven by technology evolution and
future, when the volume of B2C electronic commerce increases
consumer demand.
further and reaches a plateau, the efficiency and effectiveness of
comparison-shopping will be a major bottleneck in the effort to
Keywords achieve customer satisfaction.
Electronic agent, comparison-shopping, classification, product
comparison agent, derivative agent, Internet ecosystem A large number of comparison-shopping agents of various types
have emerged since the mid-90s to assist consumers to make
1. INTRODUCTION choice decisions in online environments [13, 15, 19, 20]. The
earliest of this breed was the bargainfinder introduced by
From the very beginning of the development of B2C electronic Anderson Consulting [3]. It allowed a shopper to enter the name
commerce, comparison-shopping has been a popular activity of a music CD and a band s/he wanted to buy, and the software
among consumers because of the characteristics of the online would then search for the selection at nine on-line stores and
environment. Compared with traditional retail stores, online return the price list to the shopper. Subsequently, many other
shopping offers potentially unlimited alternatives to satisfy the types of comparison agents emerged, such as pricescan.com,
demand of consumers. For example, Wal-mart, the biggest brick- mysimon.com, epinion.com, and bizrate.com, operating in very
and-mortar chain store, at any time has 100,000 items available on different ways to assist consumers in the same task.
the shelf in a typical Supercenter1. However, Amazon.com, the
biggest online store can offer as many as 18 million unique items2. Many researchers have contributed valuable insights about the
This assortment is further augmented if we include the items in impact of comparison-shopping agents [5, 7, 8]. However, we do
not yet have a formal classification system for identifying,
naming, and grouping these agents that often operate in different
1
ways. For example, there is a lot of confusion about even the
http://www.business2.com/articles/mag/0,1640,37760,00.html names of these agents; essentially the same comparison-shopping
2 agent has been referred to as product comparison agent, shopbot,
http://www.ug.it.usyd.edu.au/isys1003/assignments/amazoncase.d comparison-shopping agent, recommendation agent, buyer’s
oc agent as well as aggregator by various researchers [1, 5, 7, 8, 12,
16, 18, 26]. Some of these names capture only a part of the
characteristics of comparison-shopping agents. For instance, it can
be argued that a recommendation agent and an aggregator refer to
two distinct sub-groups of comparison-shopping agents that
operate differently. An aggregator focuses on collecting
quantitative objective information of products and services (like
prices and specifications) while a recommendation agent focuses
Copyright is held by the author/owner. on collecting qualitative and subjective information of products
ICEC 2003. Pittsburgh, PA
ACM 1-58113-788-5/03/09
and services (like opinions and reviews). So, the issue of
providing a comprehensive and consistent taxonomy is important

498
to both industry experts and academic researchers dealing with more related to subjective experience and operates at a lower level
comparison-shopping agents. For example, by matching the of consciousness. The rule-based reasoning operates using
current varieties of comparison-shopping agents to the cognitive abstract symbols based on logical rules that have a certain well-
processes of online decision-making, we may be able to find new specified symbolic logical structure [10, 27]. Past research
opportunities for providing better comparison-shopping services. suggests that these two modes of thinking dominate at different
stages of thinking or may work simultaneously [27].
In this paper, we use a basic classification concept in biology as
an analogy to classify these agents. If we regard the Internet as an Specifically, when consumers compare two or more items, they
ecosystem that includes various online vendors, then comparison- actually aim to differentiate them and to evaluate the differences
shopping agents fit very well in this ecosystem as a class of based on certain preferences. These tasks in the process of
herbivores that feed on plant (original data producers like online comparison use aspects of both associative reasoning and rule-
vendors) and in turn, become the target of carnivores (upstream based reasoning.
agents or meta-agents). Sitting on the top of the information food
chain are the shopping demands of online shoppers, and their Differentiation is essentially a process dominated by rule-based
click stream will in turn be the information “food” for online reasoning because it first abstracts the products in the choice set
vendors. Together, these participants – the shoppers, vendors, into symbols like price, weight, quality, etc. And then, it operates
comparison-shopping agents and upstream agents constitute the on these symbols based on certain rules, such as ‘if all other
complete ecosystem of electronic commerce world. attributes are the same, the lower priced object is preferred’ [6,
22, 23, 27].

Evaluation involves both rule-based reasoning and associative


reasoning. An evaluation outcome can be drawn either from our
past experience directly (by associative reasoning) or from
quantified experience about the product or the service (by rule-
based reasoning) or from a mix of these two strategies.

Preference Identification is mainly achieved through associative


reasoning. Whenever we have difficulty making trade-offs among
equally attractive choices based on evaluation, we may make the
decision based on a specific preference. For example, when
comparing a Dell PC and a HP PC, if we disregard small
differences between them, the brand name becomes the only
major difference. In such a case, we usually use associative
reasoning to make a choice, say, by comparing past usage
experiences on these two brands.

The goal of a formal comparison process is to pick the choice that


maximizes our utility, based on our preferences. For example, if
we want to buy a car and need to choose from two brands, a
Camry and an Accord, we will first find out the differences
between them on various aspects such as price, reputation, after
Figure 1. Internet ecosystem
sale service, etc. Then by selecting those differences that matter to
For simplification of discussion, from now on we use the term us, we begin to evaluate these differences and combine them into
PCA (acronym for Product Comparison Agent) to refer to the an aggregated utility for comparison. These differentiation and
entire class of comparison-shopping agents. Further, for evaluation processes may be repeated for other alternatives in the
clarification, we define PCA as an online intelligent software choice set, until we feel that we can make a decision based on the
application that could retrieve, aggregate and process product current evaluation result. Often, we may not find a dominant
information and present them in an appropriate format for choice even in the situation when only two alternatives remain in
choice-related decision-making by individuals [11, 19, 20]. the final consideration set if we have to trade-off between
attributes. Then, we have to make a choice based on a subjective
preference, say, that comfort is more important than price. And
2. THE ANATOMY OF COMPARISON based on that preference, we can make a decision.
Just as biologists group organisms according to their shared
characteristics, we have to first identify the shared characteristics The three aspects described above, differentiation, evaluation and
of PCAs in order to classify them. To do so, we have to focus on preference identification are integral components of the
the characteristics of comparison, or more fundamentally, the comparison process. However, they have different properties in
anatomy of comparison. terms of the cognitive processes that are involved.

Researchers have suggested that comparison is a thinking process Specifically, differentiation is objective and quantitative in nature.
that involves two modes of thinking: associative reasoning and Both the input and output of the comparison process are objective
rule-based reasoning. The associative reasoning operates by data, irrespective of differences between decision-makers. In
grouping objects with other objects that are most similar [24]. It is

499
terms of designing the PCA, it implies that the differentiation 3.1 Differentiation PCA
operation can be standardized with fixed functionalities.
Differentiation PCAs are those comparison-shopping agents that
Evaluation is a mixed process involving both objective and provide differentiation information for various attributes of choice
subjective information, so that different decision-makers may alternatives. Currently, the differentiation PCAs available are
have different evaluations of the same product. The input of mostly price comparison agents such as pricescan.com. These
evaluation can be objective data (e.g. price of product) or agents usually target 40-70 or more online vendors and establish a
quantified subjective data (e.g. quality of service). The output is formal or informal partnership with each of them.
quantified or aggregated utility information for each alternative.
In many cases, evaluation may be conducted based on common A formal partnership benefits both the agent and the vendors. On
sense. For example, if the only difference between product A and one hand, it enables the agent to extract product information from
product B is price and B is less expensive, we can evaluate B as vendors without being blocked and if necessary, to contact
better than A. However, the situation will become more complex vendors to obtain updated product catalogs directly. On the other
when the differences are not limited to one dimension, and the hand, the vendors can put restrictions on the information being
attributes involve trade-offs. For example, Product A may be obtained by the agent, for example, by providing different pricing
different from product B in two aspects: A’s price is higher than schemes to agents so that they can achieve more accurate market
B, but A’s warranty period is also longer than B. In this case, two segmentation.
consumers may evaluate the two products differently if their
subjective values and weights for the two attributes are different3. An informal partnership usually operates in the case of more
Usually, the larger the number of differences in attributes, the powerful agents like pricescan.com. Because of their sheer market
higher the degree of subjective-ness in the evaluation output. presence, the value of visibility makes many vendors willing to
open their doors to these agents, though there is no formal
Preference identification is mainly a cognitive process based on partnership between them and there is in fact, some risk of
associative reasoning. The input of this process is scenario unfavorable comparison.
information (i.e., product within a personal usage context, such as
‘a PC for my home office’) instead of product information per se. We can further classify differentiation PCA into three species
It should be noted that the preference identification stage may not based on their “living mechanism.” They are independent
be necessary in decision-making if the evaluation process could differentiation PCA, dependent differentiation PCA, and
generate a single best choice based on the utility of each embedded differentiation PCA.
alternative. Preference identification happens when the decision-
maker has trade-off difficulties among several alternatives and
cannot reach a final decision. So preference identification is
complementary to evaluation. For example, an online shopper
finds two models of Personal Digital Assistants equally attractive:
one has a larger memory and the other has a faster CPU. He feels
difficulty in making a choice between these two and checks with a
PCA site such as epinion.com and finds one piece of evaluation is
especially relevant to him: because of large memory, one Figure 2. Categorization of Differentiation PCA
commenter has said, she was able to pack more applications into
the PDA and make it really convenient for various purposes.
Because this online shopper is also planning to use the PDA for
3.1.1 Independent PCA
similar purposes, he is able to make the decision to buy the brand Independent PCAs usually do not have formal partnership
with the larger memory instead of the faster CPU. programs with vendors. Their revenues come from advertisements
of third party vendors that place ads on the search results page4,
the method currently adopted by Google.com [9]. By adopting
3. CLASSIFICATION OF PCA this method, they can remain neutral when providing
Based on the above discussion, we can group PCAs based on their differentiation information. Pricescan.com and NexTag.com are
distinction in focusing on these three aspects of comparison, typical agents in this category.
namely differentiation, evaluation and preference identification.
Specifically, we propose that there are three categories of PCAs:
3.1.2 Dependent PCA
differentiation PCA, evaluation PCA, and preference
identification PCA. Dependent PCAs are agents that do have formal partnership with
online vendors. The predecessor of a dependent PCA is the
SABER system for airline seats reservation and ticketing created
by American Airline. Initially, this system always displayed
American Airline’s ticket information at the top for any agent
3 inquiry as long as American Airline had a ticket available for that
For details: If A is $20 more expensive than B and A has a
warranty period that is 6 months longer than B, which one offers
4
more utility to consumer? Different people may have different The advertisement is not in the search result but accompany the
opinions. Some may think $20 is worth more than the additional 6 result. Usually the agent will use data mining methods to list
months warranty and hence, decide that B is a better option, while relevant advertisements that might be interesting to the consumer
others may decide the exact opposite. submitting the search query.

500
route, which put American Airline in a favorable competitive new TV serial or movie?). The unidimensional PCA that handles
position relative to other airlines. Pricewatch.com and this may collect user ratings, aggregate them and present rating
pricegrabber.com are both prominent dependent differentiation information as simple percentage number for the mobile phone
agents. They have their respective partner programs and because user, or as richer chart graph for the PC user to review.
of their increasing market power, the entry cost for vendors
wishing to participate is getting higher. Nowadays, it has already For unidimensional PCA, the immediate research task is to
become very costly for small vendors to participate in their identify and categorize the general evaluation goals of consumers.
programs. For example, do consumers need traffic route suggestions based
on travel time or travel distance? Based on such goals, PCAs can
3.1.3 Embedded PCA provide evaluation services in a predictable format.
The simplest species of differentiation PCA appears embedded in
websites of major retailers such as Amazon.com. This type of 3.2.2 Multidimensional PCA
PCA provides the differentiation information as part of the Multidimensional PCA extracts and collects quantified (subjective
product page content. It usually involves only one or two or objective) alternative information on multiple dimensions and
dimensions of differentiation (e.g., price, condition) for aggregates them for review. Multidimensional evaluation PCA is
simplification of comparison. By clicking the comparison button, already popular. Bizrate.com is one prominent agent of this type.
customer will be led to another page, which has detailed It collects customer experiences at several online stores and asks
comparison information. We call this type of PCA an embedded customers to quantify these into ratings on various dimensions,
PCA because it is embedded into the same product page for then aggregates them based on a normalized scale for
implicit differentiation. Embedded PCA is very efficient because comparison.
it only needs limited information processing capabilities on the
customer side, while at the same time it provides some
comparison information to allow customers to make more
3.3 Preference Identification PCA
informed decisions. Preference agents collect experiences of online shoppers for
different products or services and then aggregate them for
decision-making. Preference agents emerged because of the
3.2 Evaluation PCA analogical way in which human beings think and make decisions
Evaluation PCAs are comparison-shopping agents that could [21, 28]. They provide different situational contexts for the
provide quantified evaluation information on alternatives. The comparison and evaluation of products or services instead of
input information of evaluation agents can be quantified comparing products or service per se. Context information is
subjective experience (e.g. quality of service) or objective collected from peer reviewers. So, if consumers cannot find
information of alternative choices. Based on the dimensions of appropriate context-relevant information based on their own
differentiation, we have at least two species of evaluation PCA: experiences, they can use peer experiences as surrogates to
unidimensional evaluation PCA and multidimensional evaluation identify their preferences and make choice decisions. This is
PCA. particularly relevant for services (e.g., hotels, restaurants,
concerts, cruises) where preferences can vary significantly across
situational contexts (e.g., usage occasions, seasons, goals).

Epinion.com is representative of a preference PCA. At


epinion.com, online shoppers submit their opinions about
products and services. Epinion.com then aggregates them and
summarizes them for use. When an online shopper wants to
compare several options in a particular category, s/he may review
Figure 3. Categorization of Evaluation PCA opinions on these options at Epnions.com and if a piece of opinion
matches his or her usage context, it may be used to make a choice
3.2.1 Unidimensional PCA decision.
Unidimensional PCA is almost non-existent for the time being.
Products can hardly be differentiated in just one dimension Preference PCA is different from Evaluation PCA in processing
anyway. However, such a species can become very useful when subjective information. Preference PCA does not quantify
the demand for information is limited by channel capacity or subjective information while Evaluation PCA quantifies
particular consumer need. For example, with the increasing subjective evaluations or ratings into data format.
popularity of accessing Internet via mobile phone, we can expect
that a demand for simple and quick evaluation services will The following table summarizes the key aspects of the three
emerge for many needs (e.g, traffic routes, airline prices). A categories of PCAs.
prototype model may look similar to some online survey agents
that examine consumer attitudes (for example, how enjoyable is a

501
Table 1. Comparison of PCAs

Differentiation PCA Evaluation PCA Preference PCA

Input format Objective/Quantitative Subjective/Objective Subjective/Descriptive


Quantification
Processing Categorization Summarization
/Aggregation
Output format Data format Data format Text format

4. META-AGENT AND THE


INFORMATION FOOD CHAIN
In the previous section, we talked about the classification of
PCAs. If we regard the Internet as an ecosystem and comparison-
agents as species that feed on information and in turn, produce
information for others, we can predict that as the demand for
further aggregated comparisons increases, a new type of PCA, the
Meta-agent will emerge. And in addition, all these agents as well
as the online shopper demand and click-stream will form an
information food chain within the ecosystem of the Internet.

4.1 Derivative PCA


Meta-agents are agents that feed on the information produced by
other agents. The Meta-search engine, askjeeves.com is one
example of such an agent. There are vast number of Meta-agents
under development. Here, we focus on Meta-agents that feed on
PCAs, or the so-called Derivative PCA. There are hardly any
Derivative PCAs on the Internet now, so what we will do here is
establish the concept of Derivative PCA, describe their
characteristics, and demonstrate that the emergence of such agents
is inevitable.
Figure 4: Relationship between derivative PCA and PCA
Derivative PCAs are comparison-shopping agents that collect
information from other PCAs and aggregate them for use. Their
The emergence of a formal derivative PCA is perhaps
information output will be updated whenever the information
unavoidable because of current inefficiencies in the way that
from their underlying PCAs changes. For example, a consumer
PCAs operate. Many researchers have suggested that the network
wants to buy several products and send the demand to a derivative
structure of the Internet is not randomly connected but connected
PCA. The derivative PCA will probe various differentiation
in a way that follows power law distribution [2, 17]. That is to
PCAs, evaluation PCAs and preference PCAs, obtain necessary
say, a majority of the nodes have only a limited number of
information regarding these products, aggregate them, and then
connections while a few popular nodes share a huge number of
present to consumers in an appropriate format. If any information
connections. With a structure that follows power law distribution,
is updated by underlying PCAs during the process (e.g. shipping
we have the danger of missing a lot of small vendors (nodes) that
charge changes), the derivative PCA’ s information output format
have limited connections (web visibility) though they may provide
will change accordingly (e.g. total price change).
quality services. For example, search engines like Google utilize
this power law distribution structure to provide search services
Many PCAs already possess some derivative features. For
[4]. This implies that when we use Google to search for certain
example, many differentiation PCAs began to provide vendor-
products, we may invariably get directed to major branded
rating information along with the product attribute information
vendors like Amazon.com or buy.com, rather than to smaller
they collect from vendors. This rating information is obtained
vendors. Those small but quality vendors cannot be effectively
from external evaluation agents like bizrate.com or their own
selected because of their weak web visibility.
embedded evaluation agents. In either case, like the evolution of
species, these differentiation agents are no longer as simple in
The emergence of PCA only partially solved this problem: they
functioning as they used to be. They are evolving into a more
collect and aggregate information from these small vendors but at
complex species with a higher level of information extracting
most, manage to cover a small part of the vast Internet. So, we can
status by synthesizing two types of information.
expect that with the further expansion of Internet and emergence

502
of even more PCAs, derivative PCAs will inevitably emerge to advanced PCAs and online shoppers will make up an information
collect and aggregate relevant information from these PCAs and food chain within the Internet ecosystem.
naturally replace them to assist online shoppers in decision-
making.

We can also expect that derivative PCAs will emerge from more
6. REFERENCE
than one category. Some of them may focus on aggregating
information of differentiation and evaluation PCAs. They may [1] Bakos, Y. The Emerging Role of Electronic
quite probably evolve from current differentiation PCAs. Others Marketplaces on the Internet. Communications of the
may focus on collecting and aggregating evaluation and ACM,41(8), 1998, 35-42.
preference information to prevent online shopper from making [2] Barabási, A.-L. The physics of the Web. Physics World,
erroneous decision based on false preferences. The book review July 2001.
and rating system implemented by Amazon.com is one such kind [3] Belman, M.H. CommerceNet Case Study:
of agent in the making. It combines an evaluation PCA and a BargainFinder, CommerceNet, 1996.
preference PCA by allowing readers to post their comments on the [4] Brin, S. and Page, L., The Anatomy of a Large-Scale
book and at the same time rate the comments provided by others. Hypertextual Web Search Engine. source: http://www-
So, by observing the approval and disapproval rate (evaluation) on db.stanford.edu/~backrub/google.html
each comment (preference), potential book buyers will be better [5] Brynjolfsson, E. and Smith, M.D. The Great Equalizer?
informed while making their decisions. Consumer Choice Behavior at Internet Shopbots.
Working Paper.
[6] Card, S.K., Moran, T.P. and Newell, A. The psychology
4.2 Information food chain of human-computer interaction. Erlbaum, Hillsdale, NJ,
Derivative PCA is by no means the final link in the chain of 1983.
information processing in the ecosystem of the Internet. More [7] Crowston, K., Price Behavior In a Market With Internet
sophisticated comparison agents will emerge as a consequence of Buyer' s Agents. in ICIS97, (Atlanta, GA, 1997), 481-
evolving consumer demand. To better illustrate this point, we can 482.
conceptualize the entire online comparison-shopping system as an [8] Crowston, K. and MacInnes, I. The effects of market-
information food chain. There are two sources of original enabling Internet agents on competition and prices.
information: those provided by online vendors and those coming Journal of Electronic Commerce Research, 1 (4).
from online shoppers. Online vendors provide product [9] Elgin, B. and Mullaney, T.J. Search engines are picking
information and online shoppers provide click-stream, reviews, up steam BusinessWeek, 2003, 86-87.
and user information. Feeding on that category of information are [10] Fodor, J.A. and Pylyshyn, Z.W. Connectionism and
the various PCAs we discussed such as differentiation PCA, cognitive architecture: A critical analysis. Cognition,
evaluation PCA and preference PCA. Online shoppers are 28. 3-71.
currently using the information provided by these PCA, but with [11] Franklin, S. and Graesser, A., Is it an Agent, or just a
the evolution of consumer demand, derivative PCA will substitute Program?: A Taxonomy for Autonomous Agents. in the
the first-generation PCAs to provide better assistance to Third International Workshop on Agent Theories,
consumers. They will combine information from various PCA and Architectures, and Languages, (Springer-Verlag, 1996).
aggregate them for consumer. Derivative PCA may also be [12] Green, H. A Cybershopper' s Best Friend Businessweek,
replaced by more advanced PCA with the evolution of consumer 1998.
demand. For example, we may expect a more intelligent PCA will [13] Grewal, D., Monroe, K.B. and Krishnam, R. The Effects
emerge to evaluate the trustworthiness of information provided by of Price-Comparison Advertising on Buyers'
derivative PCA so consumers will feel more confident when using Perceptions of Acquisition Value, Transaction Value,
the information provided by derivative PCAs. and Behavioral Intentions. Journal of Marketing, 62.
46-59.
By including all above into one system, we will get an [14] Iyengar, S.S. and Lepper, M.R. When Choice is
information food chain for online comparison-shopping. In one Demotivating: Can One Desire Too Much of a Good
direction, information is produced, aggregated and consumed Thing? Journal of Personality and Social Psychology,
through the online vendor, PCA, derivative PCA, more advanced 79 (6).
PCA and online shopper. In the other direction, revenues [15] Jacso, P. Shopbots: shopping robots for electronic
generated by online shopper will get distributed among agents and commerce Online, 1998, 14-16.
online vendors as determined by a market mechanism. [16] Kuttner, R. The Net: A Market too Perfect for Profits
Businessweek, 1998.
[17] Lawrence, S. and Giles, C.L. Accessibility of
5. SUMMARY information on the web. Nature, 400. 107-109.
In this paper, we propose a classification schema for web-based [18] Madnick, S. and Siegel, M. Seizing the Opportunity:
comparison-shopping agents. There are three basic categories of Exploring Web Aggregation. MISQ Executive
PCA based on our classification scheme. They are differentiation (Forthcoming).
PCAs, evaluation PCAs and preference PCAs. We demonstrate [19] Maes, P. Agents that reduce work and information
that more sophisticated comparison-shopping agents, that we overload Communications of the ACM, 37(7), 1994, 30-
named derivative PCAs and more advanced PCAs, will emerge 40.
with the expansion of Internet and increasing consumer demand.
And as a result, online vendors, PCAs, derivative PCAs, more

503
[20] Mariani, M. Agent, agent! Occupational Outlook [26] Sinha, I. Cost transparency: The Net’s real threat to
Quarterly, 41 (2). 10-13. prices and brands. Harvard Business Review (March-
[21] Morowitz, H.J. and Singer, J.L. (eds.). The Mind, The April 2000). 43-50.
Brain, and Complex Adaptive Systems. Addison-Wesley [27] Sloman, S.A. Two systems of Reasoning. in Gilovich,
& Benjamin Cummings, Boston, MA, 1995. T., Griffin, D. and Kahneman, D. eds. Heuristics and
[22] Neisser, U. the multiplicity of thought. Britich Journal Biases: The Psychology of Intuitive Judgment,
of Psychology, 54. 1-14. Cambridge University Press, Cambridge, UK, 2002,
[23] Payne, J.W., Bettman, J.R. and Johnson, E.J. The 379-396.
adaptive decision maker. Cambridge University Press, [28] Streufert, S. Complexity and Complex Decision
Cambridge, 1993. Making: Convergences Between Differentiation and
[24] Rumelhart, D.E. and Zipser, D. Feature discovery by Integration Approaches to the Prediction of Task
competitive learning. Cognitive Science, 9. 75-112. Performance. Journal of Experimental Social
[25] Schwartz, B., Ward, A., Monterosso, J., Lyubomirsky, Psychology, 6. 494-509.
S., White, K. and Lehman, D.R. Maximizing Versus [29] Wan, Y., Menon, S. and Ramaprasad, A., How it
Satisficing: Happiness Is a Matter of Choice. Journal of happens: a conceptual exploration of choice overload in
Personality and Social Psychology, 83 (5). online decision-making by individuals. in AMCIS,
(Tampa, Florida, USA, 2003).

504

You might also like