Identifying key media events and modeling causal relationships between key events and reported feelings

Abstract

Example methods, systems, apparatus and machine readable media are disclosed to identify candidate media events for modification. An example method includes dividing a media instance into components and correlating physiological response data from a subject exposed to the media with the components to form correlated data. The example method also includes processing the correlated data to identify transitions representative of changes in a subject response. The example method also includes parsing the components a plurality of events based on the transitions. In addition, the example method includes identifying events of the plurality of events as candidates for modification based on the changes in the subject response.

Description

BACKGROUND

Advertisers, media producers, educators and other relevant parties have long desired to understand the responses their targets—customers, clients and pupils—have to their particular stimulus in order to tailor their information or media instances to better suit the needs of these targets and/or to increase the effectiveness of the media instance created. A key to making a high performing media instance is to make sure that every event in the media instance elicits the desired responses from the viewers, not responses very different from what the creator of the media instance expected. The media instance herein can be but is not limited to, a video, an advertisement clip, a movie, a computer application, a printed media (e.g., a magazine), a video game, a website, an online advertisement, a recorded video, a live performance, a debate, and other types of media instance from which a viewer can learn information or be emotionally impacted.

It is well established that physiological response is a valid measurement for viewers' changes in emotions and an effective media instance that connects with its audience/viewers is able to elicit the desired physiological responses from the viewers. Every media instance may have its key events/moments—moments which, if they do not evoke the intended physiological responses from the viewers, the effectiveness of the media instance may suffer significantly. For a non-limiting example, if an ad is intended to engage the viewers by making them laugh, but the viewers do not find a 2-second-long punch-line funny, such negative responses to this small piece of the ad may drive the overall reaction to the ad. Although survey questions such as “do you like this ad or not” have long been used to gather viewers' subjective reactions to a media instance, they are unable to provide more insight into why and what have caused the viewers reacted in the way they did.

SUMMARY

An approach enables an event-based framework for evaluating a media instance based on key events of the media instance. First, physiological responses are derived and aggregated from the physiological data of viewers of the media instance. The key events in the media instance can then be identified, wherein such key events drive and determine the viewers' responses to the media instance. Causal relationship between the viewers' responses to the key events and their surveyed feelings about the media instance can further be established to identify why and what might have caused the viewers to feel the way they do.

Such an approach provides information that can be leveraged by a creator of the media instance to improve the media instance. For a non-limiting example, if a joke in an advertisement is found to drive purchase intent of the product advertised, but the advertisement's target demographic does not respond to the joke, the joke can be changed so that the advertisement achieves its goal: increasing product purchase intent in the target demographic.

This Summary is provided to introduce a selection of concepts in a simplified form that are further described below in the Detailed Description. This Summary is not intended to identify key features or essential features of the claimed subject matter, nor is it intended to be used to limit the scope of the claimed subject matter. These and other advantages of the present disclosure will become apparent to those skilled in the art upon a reading of the following descriptions and a study of the several figures of the drawings.

BRIEF DESCRIPTION OF THE DRAWINGS

FIG. 1 depicts an example of a system to support identification of key events in a media instance that drive physiological responses from viewers.

FIG. 2 depicts a flowchart of an exemplary process to support identification of key events in a media instance that drive physiological responses from viewers.

FIGS. 3( a)-(c) depict exemplary traces of physiological responses measured and exemplary dividing lines of events in a media instance.

FIGS. 4( a)-(c) depict exemplary event identification results based on different event defining approaches.

FIG. 5 depicts results from exemplary multivariate regression runs on events in an advertisement to determine which events drive the viewers' responses the most.

FIGS. 6( a)-(b) depict exemplary correlations between physiological responses from viewers to key jokes in an ad and the surveyed intent of the viewers to tell others about the ad.

DETAILED DESCRIPTION

The disclosure is illustrated by way of example and not by way of limitation in the figures of the accompanying drawings in which like references indicate similar elements. It should be noted that references to “an” or “one” or “some” example(s) in this disclosure are not necessarily to the same example, and such references mean at least one. Although the subject matter is described in language specific to structural features and/or methodological acts, it is to be understood that the subject matter defined in the appended claims is not necessarily limited to the specific features or acts described above. Rather, the specific features and acts described above are disclosed as example forms of implementing the claims.

Although the diagrams depict components as functionally separate, such depiction is merely for illustrative purposes. It will be apparent to those skilled in the art that the components portrayed in this figure can be arbitrarily combined or divided into separate software, firmware and/or hardware components. Furthermore, it will also be apparent to those skilled in the art that such components, regardless of how they are combined or divided, can execute on the same computing device or multiple computing devices, and wherein the multiple computing devices can be connected by one or more networks.

Physiological data, which includes but is not limited to heart rate, brain waves, electroencephalogram (EEG) signals, blink rate, breathing, motion, muscle movement, galvanic skin response and any other response correlated with changes in emotion of a viewer of a media instance, can give a trace (e.g., a line drawn by a recording instrument) of the viewer's responses while he/she is watching the media instance. The physiological data can be measure by one or more physiological sensors, each of which can be but is not limited to, an electroencephalogram, an accelerometer, a blood oxygen sensor, a galvanometer, an electromyograph, skin temperature sensor, breathing sensor, and any other physiological sensor.

The physiological data in the human body of a viewer has been shown to correlate with the viewer's change in emotions. Thus, from the measured “low level” physiological data, “high level” (i.e., easier to understand, intuitive to look at) physiological responses from the viewers of the media instance can be created. An effective media instance that connects with its audience/viewers is able to elicit the desired emotional response. Here, the high level physiological responses include, but are not limited to, liking (valence)—positive/negative responses to events in the media instance, intent to purchase or recall, emotional engagement in the media instance, thinking—amount of thoughts and/or immersion in the experience of the media instance, adrenaline—anger, distraction, frustration, and other emotional experiences to events in the media instance. In addition, the physiological responses may also include responses to other types of sensory stimulations, such as taste and/or smell, if the subject matter is food or a scented product instead of a media instance.

FIG. 1 depicts an example of a system 100 to support identification of key events in a media instance that drives physiological responses from viewers. In the example of FIG. 1, the system 100 includes a response module 102, an event defining module 104, a key event module 106, and a reaction database 108.

The response module 102 is a software component which while in operation, first accepts and/or records physiological data from each of a plurality of viewers watching a media instance, then derives and aggregates physiological responses from the collected physiological data. Such derivation can be accomplished via a plurality of statistical measures, which include but are not limited to, average value, deviation from mean, 1st order derivative of the average value, 2nd order derivative of the average value, coherence, positive response, negative response, etc., using the physiological data of the viewers as inputs. Facial expression recognition, “knob” and other measures of emotion can also be used as inputs with comparable validity. Here, the physiological data may be either be retrieved from a storage device or measured via one or more physiological sensors, each of which can be but is not limited to, an electroencephalogram, an accelerometer, a blood oxygen sensor, a galvanometer, an electromyograph, and any other physiological sensor either in separate or integrated form. The derived physiological responses can then be aggregated over the plurality of viewers watching one or more media instances.

The event defining module 104 is a software component which while in operation, defines and marks occurrences and durations of a plurality of events happening in the media instance. The duration of each of event in the media instance can be constant, non-linear, or semi-linear in time. Such event definition may happen either before or after the physiological data of the plurality of viewers has been measured, where in the later case, the media instance can be defined into the plurality of events based on the physiological data measured from the plurality of viewers.

The key event module 106 is a software component which while in operation, identifies one or more key events in the media instance and reports the key events to an interested party of the media instance, wherein the key events drive and determine the viewers' physiological responses to the media instance. Key events in the media instance can be used to pinpoint whether and/or which part of the media instance need to be improved or changed, and which part of the media instance should be kept intact. For non-limiting examples, the key event module may identify which key event(s) in the media instance trigger the most positive or negative responses from the viewers, or alternatively, which key event(s) are polarizing events, e.g., they cause large discrepancies in the physiological responses from different demographic groups of viewers, such as between groups of men and women, when the groups are defined by demographic characteristics. In addition, the key event module is operable to establish a causal relationship between the viewers' responses to the events in the media instance and their surveyed feelings about the media instance so that creator of the media instance may gain insight into the reason why and what key events might have caused the viewers to feel the way they do.

The reaction database 108 stores pertinent data of the media instance the viewers are watching, wherein the pertinent data includes but is not limited to survey questions and results asked for each of the plurality of viewers before, during, and/or after their viewing of the media instance. In addition, the pertinent data may also include but is not limited to the following:

• • Events/moments break down of the media instance;
  • Key events in the media instance;
  • Metadata of the media instance, which can include but is not limited to, production company, brand, product name, category (for non-limiting examples, alcoholic beverages, automobiles, etc), year produced, target demographic (for non-limiting examples, age, gender, income, etc) of the media instance.
  • If the subject matter is food or a scented product instead of a media instance, the surveyed reactions to the taste or smell of a key ingredient in the food or scented product. Here, the term database is used broadly to include any known or convenient means for storing data, whether centralized or distributed, relational or otherwise.

While the system 100 depicted in FIG. 1 is in operation, the response module 102 derives aggregated physiological responses from the physiological data of a plurality of viewers watching a media instance. The key event module 106 identifies, among the plurality of events in the media instance as defined by the event defining module 104, one or more key events that drive and determine the viewers' physiological responses to the media instance based on the aggregated physiological responses from the viewers. In addition, the key event module 106 may retrieve outcomes to questions surveyed from the viewers of the media instance from the reaction database 108, and correlates the viewers' responses to the key events and their surveyed feelings about the media instance to determine what might have caused the viewers to feel the way they do. The entire approach can also be automated as each step of the approach can be processed by a computing device, allowing for objective measure of a media without much human input or intervention.

FIG. 2 depicts a flowchart of an exemplary process to support identification of key events in a media instance that drive physiological responses from viewers. Although this figure depicts functional steps in a particular order for purposes of illustration, the process is not limited to any particular order or arrangement of steps. One skilled in the art will appreciate that the various steps portrayed in this figure could be omitted, rearranged, combined and/or adapted in various ways.

Referring to FIG. 2, physiological responses can be derived and aggregated from the physiological data of a plurality of viewers watching a media instance at block 202. At block 204, the media instance can be defined into a plurality of events, and correlation between the physiological responses from the viewers to the key events in the media instance and a surveyed outcome of their feelings about the media instance can optionally be established at block 206. At block 208, key events in the media instance can be identified based on the aggregated physiological responses from the viewers and/or the correlation between the physiological responses and a surveyed outcome. Finally, the key events and/or their correlation with the surveyed outcome are reported to an interested party of the media instance at block 210, wherein the interested party may then improve the media instance based on the key events and/or the correlations.

Events Definition

In some examples, the event defining module 104 is operable to define occurrence and duration of events in the media instance based on salient positions identified in the media instance. Once salient positions in the media instance are identified, the events corresponding to the salient positions can be extracted. For a non-limiting example, an event in a video game may be defined as a “battle tank” appearing in the player's screen and lasting as long as it remains on the screen. For another non-limiting example, an event in a movie may be defined as occurring every time a joke is made. While defining humor is difficult, punch line events that are unexpected, absurd, and comically exaggerated often qualify as joke events.

FIG. 3( a) shows an exemplary trace of the physiological response—“Engagement” for a player playing Call of Duty 3 on the Xbox 360. The trace is a time series, with the beginning of the session on the left and the end on the right. Two event instances 301 and 302 are circled, where 301 on the left shows low “Engagement” during a game play that happens during a boring tutorial section. 302 shows a high “Engagement” section that has been recorded when the player experiences the first battle of the game. FIG. 3(b) shows exemplary vertical lines that divide a piece of media instance into many events defining every important thing that a player of the video game or other media may encounter and/or interact with.

In some examples, the event defining module 104 is operable to define occurrence and duration of events in the media instance via at least the one or more of the following approaches. The events so identified by the event defining module 104 are then provided to the key event module 106 to test for “significance” as key events in the media instance as described below.

The hypothesis approach, which utilizes human hypothesis to identify events in the media instance, wherein such events shall be tested for significance as key events.

The small pieces or time shift approach, which breaks the media instance into small pieces in time, and scans each small piece for significant switch in the viewers' responses, wherein consecutive significant small pieces can be integrated as one key event. For the non-limiting example of FIG. 4(a), the small pieces are each ⅕ second in length and consecutive small pieces that are found to be significant indicate an event, such as 401 and 402. For the exemplary car ad shown in FIG. 4(b), 403 represents the first 10 seconds of the car ad as a cross-component event, and 404 represents a cross-ad music event.

The turning point approach, which finds where the aggregated physiological responses (traces), first derivative, and second derivative of aggregated trace(s) have roots and uses them as possible event cut points (delimiters). Here, roots of the aggregate traces can be interpreted as points when the viewers' aggregated physiological responses transition from above average to below average, or from positive to negative. Roots in the first derivative of the aggregate traces can be interpreted as ‘turning points’, at which the physiological responses transition from a state of increasing positivity to increasing negativity, or vice versa. Roots in the second derivative of the aggregate traces can also be interpreted as ‘turning points’, points, at which the physiological responses begin to slow down the rate of increase in positivity. All such roots are then collected in a set s. For every pair i,j of roots in the set s for which j occurs after i in the media instance, the event which starts at i and ends at j is tested for significance as a key event. Note here that i and j do not have to be consecutive in time.

The multi-component event approach, which breaks the media instance down into components and then divides each component into events. A media instance typically has many components. For a non-limiting example, an advertisement can have one or more of: voiceover, music, branding, and visual components. All points in the media instance for which there is a significant change in one of the components, such as when the voiceover starts and ends, can be human marked. As with the turning point approach, all the marked points can be collected in the set s. For every pair i,j of roots in the set s for which j occurs after i in the media instance, the event which starts at i and ends at j is tested for significance as a key event. While this approach requires greater initial human input, it may provide more precise, more robust results based on automated higher-level analysis and the benefits would outweigh the costs. For a non-limiting example, a car ad can be broken down into visual, dialogue, music, text, and branding components, each with one or more events. For the exemplary car ad shown in FIG. 4(c), 405 represents a visual event, 406 represents a dialogue event, and 407 represents a music event.

Key Events Identification

In some examples, the key event module 106 is operable to accept the events defined by the event defining module 104 and automatically spot statistically significant/important points in the aggregated physiological responses from the viewers relevant to identify the key moments/events in the media instance. More specifically, the key event module is operable to determine one or more of:

• • if an event polarizes the viewers, i.e., the physiological responses from the viewers are either strongly positive or strongly negative.
  • if the physiological responses vary significantly by a demographic factor.
  • if the physiological responses are significantly correlated with the survey results.
  • if an event ranks outstandingly high or low compared to similar events in other media instances For a non-limiting example, FIG. 3(c) shows two exemplary traces of the “Engagement” response of a video game player where the boxes 303, 304, and 305 in the pictures correspond to “weapon use” events. At each point where the events appear, “Engagement” rises sharply, indicating that the events are key events for the video game.

In some examples, the key events found can be used to improve the media instance. Here “improving the media instance” can be defined as, but is not limited to, changing the media instance so that it is more likely to achieve the goals of the interested party or creator of the media instance.

In some examples, the key event module 106 is further operable to establish a casual relationship between surveyed feelings about the media instance and the key events identified based on the physiological responses from the viewers. In other words, it establishes a correlation between the physiological responses from the viewers to key events in the media instance and a surveyed outcome, i.e., the viewers' reported feelings on a survey, and reports to the interested parties (e.g. creator of the event) which key events in the media instance actually caused the outcome. Here, the outcome can include but is not limited to, liking, effectiveness, purchase intent, post viewing product selection, etc. For a non-limiting example, if the viewers indicate on a survey that they did not like the media instance, something about the media instance might have caused them to feel this way. While the cause may be a reaction to the media instance in general, it can often be pinned down to a reaction to one or more key events in the media instance as discussed above. The established casual relationship explains why the viewers report on the survey their general feelings about the media instance the way they do without human input.

In some examples, the key event module 106 is operable to adopt multivariate regression analysis via a multivariate model that incorporates the physiological responses from the viewers as well as the surveyed feelings from the viewers to determine which events, on average, are key events in driving reported feelings (surveyed outcome) about the media instance. Here, the multivariate regression analysis examines the relationship among many factors (the independent variables) and a single, dependent variable, which variation is thought to be at least partially explained by the independent variables. For a non-limiting example, the amount of rain that falls on a given day varies, so there is variation in daily rainfall. Both the humidity in the air and the number of clouds in the sky on a given day can be hypothesized to explain this variation in daily rainfall. This hypothesis can be tested via multivariate regression, with daily rainfall as the dependent variable, and humidity and number of clouds as independent variables.

In some examples, the multivariate model may have each individual viewer's reactions to certain key events in the media instance as independent variables and their reported feeling about the media instance as the dependent variable. The coefficients from regressing the independent variables on the dependent variable would determine which key events are causing the reported feelings. Such a multivariate model could be adopted here to determine what set of key events most strongly affect reported feelings from the viewers about the media instance, such as a joke in an advertisement. One characterization of such event(s) is that the more positive (negative) the viewers respond to the event(s), the more likely the viewers were to express positive feelings about the media instance. For a non-limiting example, a multivariate regression can be run on multiples events (1, 2 . . . n) within an entire montage sequence of an advertisement to determine which events drive liking the most, using relationship between reported feelings about the ad and the emotional responses from the viewers to the events in the ad as input. The results of the multivariate regression runs shown in FIG. 5 indicate that 2 out of the 6 events tested in the ad drive the viewers' responses the most, while the other 4 events do not meet the threshold for explanatory power.

In an automated process, this multivariate regression may be run stepwise, which essentially tries various combinations of independent variables, determining which combination has the strongest explanatory power. This is a step toward creating the causal relationship between the viewers' responses to the events and their surveyed feelings about the media instance. For a non-limiting example, if response to joke #2 is correlated with indicated intent to purchase when holding genders and responses to jokes #1 and #3 constant, a causal conclusion can be made that joke #2 triggers the viewers' intent to purchase.

In some examples, the key event module 106 identifies the key polarizing event(s) that cause statistically significant difference in the surveyed outcome from different demographic groups of viewers and provides insight into, for non-limiting examples, why women do not like the show or which issue actually divides people in a political debate. The key event module 106 may collect demographic data from overall population of the viewers and categorize them into groups to differentiate the responses for the subset, wherein the viewers can be grouped one or more of: race, gender, age, education, demographics, income, buying habits, intent to purchase, and intent to tell. Such grouping information can be included in the regressions to determine how different groups report different reactions to the media instance in the survey. Furthermore, grouping/event response interaction variables can be included to determine how different groups respond differently to the key events in the media instance. For key events that are polarizing, demographic information and/or interaction variables of the viewers can also be included to the multivariate model capture the combined effect of the demographic factor and the reaction to the polarizing key events.

For a non-limiting example, the viewers of an ad can be first asked a survey question, “How likely are you to tell someone about this particular commercial—meaning tell a friend about this ad you've just seen” as shown in FIG. 6(a). The viewers of the ad are broken into two groups based on indicated likelihood to tell someone about the commercial—the affirmative group and the negative group, and it is assumed that the viewers in both groups are normally distributed with the same variance. The emotional responses from the viewers in the groups to two key jokes 601 and 602 in the ad are then compared to their surveyed reactions to test the following hypothesis—“the group that indicated they were likely to tell someone will have a stronger positive physiological response to the two key jokes than the group that indicated they were unlikely to tell someone.” The affirmative group that indicated they were likely to tell a friend had, on average, a more positive reaction to both jokes than the negative group that indicated they were unlikely to tell a friend. In both cases, experiments using the change in individual liking as the metric to measure the physiological response rejects the null hypothesis—that there was no difference in emotional response to the jokes between the two groups—at above the 95% confidence level. Referring to graphs in FIG. 6(a), the X axis displays the testers' likelihood to tell a friend as indicated on the post-exposure survey and the Y axis displays the testers' emotional response to the joke. The triangles represent the “Go” people—those who indicated they were likely (to varying degrees) to tell their friend about the spot. The circles represent those who indicated that they were unlikely to do so. Note the upward trends—the more positive the emotional reaction to the joke, the greater indicated likelihood to tell a friend about the spot. FIG. 6(b) further summarizes the physiological responses to both jokes, where the reaction to the Crying Wife is on the X axis and the reaction to the Dead Husband is on the Y axis. An imaginary line around is drawn around the viewers who reacted relatively positively to both jokes. Note that this line corrals most of the black diamonds, which represent viewers who indicated they would tell a friend about the ad.

One example may be implemented using a conventional general purpose or a specialized digital computer or microprocessor(s) programmed according to the teachings of the present disclosure, as will be apparent to those skilled in the computer art. Appropriate software coding can readily be prepared by skilled programmers based on the teachings of the present disclosure, as will be apparent to those skilled in the software art. The disclosure may also be implemented by the preparation of integrated circuits or by interconnecting an appropriate network of conventional component circuits, as will be readily apparent to those skilled in the art.

One example includes a computer program product which is a machine readable medium (media) having instructions stored thereon/in which can be used to program one or more computing devices to perform any of the features presented herein. The machine readable medium can include, but is not limited to, one or more types of disks including floppy disks, optical discs, DVD, CD-ROMs, micro drive, and magneto-optical disks, ROMs, RAMs, EPROMs, EEPROMs, DRAMs, VRAMs, flash memory devices, magnetic or optical cards, nanosystems (including molecular memory ICs), or any type of media or device suitable for storing instructions and/or data. Stored on any one of the computer readable medium (media), the present disclosure includes software for controlling both the hardware of the general purpose/specialized computer or microprocessor, and for enabling the computer or microprocessor to interact with a human viewer or other mechanism utilizing the results of the present disclosure. Such software may include, but is not limited to, device drivers, operating systems, execution environments/containers, and applications.

The foregoing disclosed examples has been provided for the purposes of illustration and description. It is not intended to be exhaustive or to limit the disclosure to the precise forms disclosed. Many modifications and variations will be apparent to the practitioner skilled in the art. Particularly, while the concept “module” is used in the examples of the systems and methods described above, it will be evident that such concept can be interchangeably used with equivalent concepts such as, class, method, type, interface, bean, component, object model, and other suitable concepts. Examples were chosen and described in order to best describe the principles of the disclosure and its practical application, thereby enabling others skilled in the art to understand the disclosure, the various examples and with various modifications that are suited to the particular use contemplated. It is intended that the scope of the disclosure be defined by the following claims and their equivalents.

Claims

1. A method comprising: dividing a media instance into a first component and a second component, the first component and second component concurrently presented;correlating first physiological response data from a first subject exposed to media with the first component and the second component to form first correlated data and second correlated data;processing, using a processor, the first correlated data to identify a first transition representative of a first change;processing, using the processor, the second correlated data to identify a second transition representative of a second change;parsing the first component into a first plurality of events based on the first transition;parsing the second component into a second plurality of events based on the second transition;identifying a first event of the first plurality of events as a first candidate for modification based on the first change; andidentifying a second event of the second plurality of events as a second candidate for modification based on the second change.

2. The method of claim 1, wherein the first component and the second component comprise at least one of voiceover, music, branding, dialogue, text, or a visual component, the first component being different from the second component.

3. The method of claim 1, wherein dividing the media instance into a first component and a second component comprises identifying a first duration of the first component and a second duration of the second component.

4. The method of claim 3, wherein the first duration and the second duration are non-linear.

5. The method of claim 1, wherein the first transition is representative of the first change between a positive response and a negative response for the first component and the second transition is representative of the second change between a positive response and a negative response for the second component.

6. The method of claim 1, further comprising correlating second physiological response data from a second subject exposed to the media with the first component and the second component to form third correlated data and fourth correlated data respectively; andprocessing, using the processor, the third correlated data to identify a third transition representative of a third change;processing, using the processor, the fourth correlated data to identify a fourth transition representative of a fourth change;parsing the first component into the first plurality of events based on the first transition and the third transition;parsing the second component into the second plurality of events based on the second transition and the fourth transition;identifying the first candidate for modification based on one or more of the first change or the third change; andidentifying the second candidate for modification based on one or more of the second change or the fourth change.

7. The method of claim 1, further comprising correlating second physiological response data from a second subject exposed to the media with the first component and the second component to form third correlated data and fourth correlated data respectively;processing the third correlated data to identify a third transition representative of a third change;processing the fourth correlated data to identify a fourth transition representative of a fourth change;parsing the first component into a third plurality of events based on the third transition;parsing the second component into a fourth plurality of events based on the fourth transition;identifying a third event of the third plurality of events as a third candidate for modification based on the third change; andidentifying a fourth event of the fourth plurality of events as a fourth candidate for modification based on the fourth change.

8. The method of claim 7, further comprising: obtaining survey data from the first subject and the second subject; andperforming a multivariant regression analysis based on the first physiological response data, the second physiological response data, and the survey data to determine which of at least of one of the first event, the second event, the third event, or the fourth event drive the survey data.

9. The method of claim 1, further comprising correlating second physiological response data from a second subject exposed to the media with the first component and the second component to form third correlated data and fourth correlated data respectively;performing a first comparison of the first correlated data and the third correlated data;performing a second comparison of the second correlated data and the fourth correlated data;detecting a discrepancy between the first response data and the second response data based on at least one of the first comparison or the second comparison, the discrepancy being indicative of a polarizing event; andidentifying the polarizing event for modification.

10. A system comprising: an analyzer to: divide a media instance into a first component and a second component, the first component and second component concurrently presented;correlate first physiological response data from a first subject exposed to media with the first component to form first correlated data;correlate the first physiological response data with the second component to form second correlated data;process the first correlated data to identify a first transition representative of a first change;process the second correlated data to identify a second transition representative of a second change;parse the first component into a first plurality of events based on the first transition;parse the second component into a second plurality of events based on the second transition;identify an event of the first plurality of events as a first candidate for modification based on the first change;identify an event of the second plurality of events as a second candidate for modification based on the second change; andan output to provide a first identity of the first candidate and a second identity of the second candidate.

11. The system of claim 10, wherein the analyzer is to divide the media instance into a first component and a second component by identifying a first duration of the first component and a second duration of the second component.

12. The system of claim 10, wherein the first transition is representative of the first change between a positive response and a negative response for the first component and the second transition is representative of the second change between a positive response and a negative response for the second component.

13. The system of claim 10, wherein the analyzer is to: correlate second physiological response data from a second subject exposed to the media with the first component to form third correlated data;correlate the second physiological response data with the second component to form fourth correlated data;process the third correlated data to identify a third transition representative of a third change;process the fourth correlated data to identify a fourth transition representative of a fourth change;parse the first component into the first plurality of events based on the first transition and the third transition;parse the second component into the second plurality of events based on the second transition and the fourth transition;identify the first candidate for modification based on one or more of the first change or the third change; andidentify the second candidate for modification based on one or more of the second change or the fourth change.

14. The system of claim 10, wherein the analyzer is to: correlate second physiological response data from a second subject exposed to the media with the first component to form third correlated data;correlate the second physiological response data with the second component to form fourth correlated data;process the third correlated data to identify a third transition representative of a third change;process the fourth correlated data to identify a fourth transition representative of a fourth change;parse the first component into a third plurality of events based on the third transition;parse the second component into a fourth plurality of events based on the fourth transition;identify a third event of the third plurality of events as a third candidate for modification based on the third change; andidentify a fourth event of the fourth plurality of events as a fourth candidate for modification based on the fourth change.

15. The system of claim 14, wherein the processor is to perform a multivariant regression analysis based on the first physiological response data, the second physiological response data, and survey data obtained from the first subject and the second subject to determine which of at least of one of the first event, the second event, the third event, or the fourth event drives the survey data.

16. The system of claim 10, wherein the analyzer is to: correlate second physiological response data from a second subject exposed to the media with the first component to form third correlated data;correlate the second physiological response data with the second component to form fourth correlated data;perform a first comparison of the first correlated data and the third correlated data;perform a second comparison the second correlated data and the fourth correlated data;detect a discrepancy between the first response data and the second response data based on at least one of the first comparison or the second comparison, the discrepancy being indicative of a polarizing event; andidentify the polarizing event for modification.

17. A tangible machine readable storage device or storage disc comprising instructions, which when executed, cause a machine to at least: divide a media instance into a first component and a second component, the first component and second component concurrently presented;correlate first physiological response data from a first subject exposed to media with the first component to form first correlated data;correlate the first physiological response data with the second component to form second correlated data;process the first correlated data to identify a first transition representative of a first change;process the second correlated data to identify a second transition representative of a second change;parse the first component into a first plurality of events based on the first transition;parse the second component into a second plurality of events based on the second transition;identify an event of the first plurality of events as a first candidate for modification based on the first change; andidentify an event of the second plurality of events as a second candidate for modification based on the second change.

18. The storage device or storage disc of claim 17, wherein the instructions cause the machine to divide the media instance into a first component and a second component by causing the machine to identify a first duration of the first component and a second duration of the second component.

19. The storage device or storage disc of claim 17, wherein the first transition is representative of the first change between a positive response and a negative response for the first component and the second transition is representative of the second change between a positive response and a negative response for the second component.

20. The storage device or storage disc of claim 17, wherein the instructions cause the machine to: correlate second physiological response data from a second subject exposed to the media with the first component to form third correlated data;correlate the second physiological response data with the second component to form fourth correlated data;perform a first comparison of the first correlated data and the third correlated data;perform a second comparison the second correlated data and the fourth correlated data;detect a discrepancy between the first response data and the second response data based on at least one of the first comparison or the second comparison, the discrepancy being indicative of a polarizing event; andidentify the polarizing event for modification.