A Cultural Market Model

Social interaction is an inevitable aspect of our lives and a very strong ingredient of our decision processes. Most of our decisions depend, at least partially, on what other people think and how they behave. The extent of the society's influence on our behaviors may range from daily decisions such as what to wear at work to political decisions such as which party to vote for in the elections. The importance of the social interaction has been reflected in the social sciences for many decades and a growing body of research continues on the intersection of various disciplines including but not limited to sociology, cognitive sciences, physics and economics. 1,2,3,4,5,6 Statistical physics has a long history of dealing with interacting particles and emergent phenomena. Some of the techniques employed by statistical physics are also applied to human populations successfully and offer us new ways to explore the dynamics in social systems. 7,8,9,10,11,12 The interaction between the macro dynamics and individual decision processes is an active area of research in computational sociology and statistical physics (i.e. sociophysics). Problems in which the agents in a community are faced with a binary decision such as to vote for or against a legislation or to buy a particular product or not are extensively studied by the statistical physicists. 2,3,8,10, 12 The Random Field Ising Model (RFIM) is a commonly investigated model to analyze such situations. 8,10,11,13,14,15 There are also other problems in which the opinions of the agents are represented as multidimensional continuous vectors and are influenced by the opinions of other agents. 16 In this paper, we investigate the effect of social interactions on the consumption behaviors of people in a market. In a cultural market, there are many items competing with each other and the decision to consume an item or not is not independent of the consumption decisions of other items. The consumers have to pick one or more items from a wider pool of opportunities.

An empirical study carried out by Salganik et al. (2006) provides experimental evidence that social influence has an effect on the consumption decisions of people. 17 In the experiment, the subjects are faced with a web based application in which they can listen to and rate as many songs as they like among 48 songs of previously unknown bands. After they listen to and rate a song, they are offered the opportunity to download the song. The study reports the results of two different experimental conditions. The first one is called the independent condition in which the subjects only see the names of the songs without any other information and make their decisions independently from the other subjects. The second condition is called the social influence condition and in this case, the number of people who have downloaded each song so far is also given to the subjects. The number of downloads of a song is called the success of the song. Any significant difference in the success of the same song between the two cases can be attributed to the availability of social information since there is no other experimental difference between the two settings.

The key finding of the study is that the availability of the social information significantly affects the way people behave. In the social influence condition, variation of the success outcomes of the songs are much higher than it is in the independent condition. This suggests that in the social influence case, some songs are downloaded many more times than they are in the independent case. Another measure they report is called the unpredictability of a song and is found by calculating the average difference between the success values of a song over different realizations of the same condition (i.e. the experimental condition is repeated several times with different subjects). If a song tends to get the same outcomes over different realizations then its unpredictability value is low otherwise it is high. As a result of the experiment, the social influence condition leads to higher unpredictability values for the songs. Borghesi and Bouchaud (2007) propose a generalization of the RFIM such that it allows to study multiple choices made by the agents simultaneously. 18 They show that it is possible to estimate some parameters of the model from the empirical A Cultural Market Model 3 data of Salganik et al. 17 . Furthermore, they report that the behavior of the model changes qualitatively for low and high values of social pressure which is in alignment with the empirical findings.

Another line of research is carried out under social percolation models. 19 In this approach, consumers and producers are modeled as adaptive agents and the hit or failure of products are studied as a result of the information contagion between the consumers.

The kind of market we are interested in is one where the items are easily reproducible so that their supply is practically unlimited and the

…(Full text truncated)…

This content is AI-processed based on ArXiv data.