Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Images, Art & Video - Orientation #38

Open
HyunkuKwon opened this issue Apr 7, 2020 · 15 comments
Open

Images, Art & Video - Orientation #38

HyunkuKwon opened this issue Apr 7, 2020 · 15 comments

Comments

@HyunkuKwon
Copy link

HyunkuKwon commented Apr 7, 2020

Post questions about the following orienting reading:

Collins, Randall. 2009. “The Micro-sociology of Violent Confrontations” and “Confrontational Tension and Incompetent Violence” (beginning of Chapter 2) from Violence: A Microsociological Theory: 37-43.

@nwrim
Copy link

nwrim commented May 22, 2020

What an interesting, wholesome reading! I will have to think more about whether I agree that violence is more of a situational process, but this gave me a lot to think about.

There are so many things to think about in this reading, but I think the point most relevant to our class is that texts, regardless of whether presented orally or written down, can be and will be distorted - alas,

when participants talk about violent situations, they tend to give a very truncated, and by their own lights, idealized version of what went on. (p.4)

Thus, the author implies that perhaps using images and videos could be a good way to supplement this distortion of events through various factors (such as myth related to violence) - to quote some parts where I got that impression,

Visual evidence shows us something about violence that we are not prepared to see. The pattern looks much the same in a wide range of incidents, in many different ethnic combinations within and across ethnic group lines (p.4)

and

The emotional pattern comes out when we see what combat actually looks like and attempt to analyze its nonverbal expressions (p.42)

Although I might be misreading the author's intention (I do not think he or she pointed this out explicitly), this gave me an interesting question - wouldn't there be a systemic bias of some sort when we analyze image and video? I know that a new era has dawned upon us and the Gen Zs and whatever younger generation than that records and pictures virtually everything, but I still think that there could be a bias in what gets recorded and what does not get recorded. I guess this bias will be prevalent in any kind of anecdotal data acquisition, but somehow I feel it will be worse on images and videos since while it is possible to (try to) remember certain events, it is impossible to go back in time and record the thing we want to see.

I feel like I am just rambling random thoughts here, but could you share some thoughts on this matter (the bias in what gets recorded and what gets unrecorded, potentially causing over-representation of what got recorded)

@wanitchayap
Copy link

I think @nwrim 's question is really important. On top of the selection bias from recording or not recording something, I think that the platform that the researchers choose to/have accessibility to collect pictures and videos from could heavily confound what are the collected contents. For example, pictures/videos on Instagram tend to be for showing the lifestyle more than just any random pictures or videos. On the other hand, pictures/videos on Twitter and Facebook have different purposes. In addition to the platform itself, these days, there are so many ways to edit photos and videos that could entirely change the meaning/feeling of certain pictures/videos. How could we as a researcher avoid these problems in observational studies?

@Yilun0221
Copy link

The most inspiring thing in this reading to me is that the authors linked biological revolution with social revolution, since this perspective sheds new light on my understanding of human behavior. My question is, since societal interactions and changes are more complex than biological ones because societal interactions are not so objective as nature science events, how can we describe, analyze and predict societal interactions or human behavior with computational methods (like machine learning methods)?

@iarakshana
Copy link

Interesting theory that violence is a situational process! The author also notes that human observation, reconstruction and recordings are the three types of ways to approach situational violence - he then goes on to imply that observation and reconstruction tend to be more reliable? over recordings but I think this might be a slightly dated opinion perhaps? He even says that photos tend be more useful than video because they can be used to closely examine emotion. The text was published in 2009 so do you think this still applies in the current day and age given a lot of advancement in technology?

@linghui-wu
Copy link

This is an interesting reading that provides theoretical foundations for a novel application of computer visions. However, due to my very limited exposure to CV, I cannot quite understand why the author claimed that the pictures can be more useful than videos, especially in the field of closely examining emotions. So I wonder if there is some new advancement in video understanding nowadays that facilitates the face recognition studies.

Moreover, I noticed that the authors delved more into how violence proceeds and how the myths on violence evolved and can be incorporated into machine learning technologies. However, in real life, there might be other non-interpersonal forms of violence that are not detailed explained in the chapters, such as tensions in diplomatic contexts or non-violence resistance. Probably I digress from the main topic but it would be great if someone could provide me any thoughts on studies that combine the implicit violence and statistical causal inference.

@timqzhang
Copy link

It is an exciting paper to discussion, and I actually get notice of one option on the video and static photograph, which echoes the question of @linghui-wu :

Still photos are often better than videos for capturing the emotional aspects of violent interaction. When we analyze a video of a conflict sequence (or indeed any video of interaction), we may slow it down to segments of micro-seconds (frame-by-frame in older camera film) to pull out just those details of bodily posture, facial expression, and sequence of micro-movements. In depictions of riots, which I use extensively in this work, still photos dramatically show the division between the active few on the violent front and the supporting mass of demonstrators.

It seems counterintuitive to think that still photos are better than videos to study the violence emotions. As mentioned in the quotation, videos can show the facial expression, micro-movements and other postures, which I personally think are far more advantageous than the listed pros of still photos, mainly in terms of the accuracy. Actually in essences, the videos consist of multiple still pictures, and why does author insist that the still photos could be better than the micro-movements? is it due to the technology progress on video processing in 2009?

@tianyueniu
Copy link

This is a really interesting sociology reading. The author mentioned in the article that “The appropriate relationship of micro- and macro-sociology is not to reduce one to the other, but to coordinate the two levels of analysis where it leads to some useful result.” My question is similar to @Yilun0221 , in cases similar to violence, context interacts with how things will develop over time, how can we build more accurate model to capture the 'randomness' in social science fields?

@WMhYang
Copy link

WMhYang commented May 29, 2020

I find the idea that we should focus on the violent situations interesting and essential as it helps eliminate statistical discrimination. In addition, when we analysis the situations, it is important that we combine micro and macro explanations together and this is where my question comes up. From my point of view, analyzing photos, videos, or even texts would be more likely to give us micro-level evidences, especially when we do not have data that go over long enough periods. In this case, is it possible to build a model (like machine learning model) to predict or impute the trend so that macro explanations could be generated?

@DSharm
Copy link

DSharm commented May 29, 2020

Interesting read for sure. I have a question slightly related to the reading but more generally related to the analysis of violence through videos, photos, audio. There's been a recent literature discussing the idea that body-cam footage of police officers interacting with civilians may not be the cure-all many had hoped it would be - in terms of objectively documenting appropriate and inappropriate police behavior - because many individuals who view such footage tend to view it with their own biases. A study tracked the eye movements of participants watching body-cam footage and found that one's underlying values/prejudices/biases determined where an individual focused more attention, and consequently whether or not they found the police behavior to be appropriate or excessive.

In this reading, the author emphasizes how the "trained eye" of the micro-sociologist can focus on the micro-movements, tensing of muscles, etc. to analyze these situations. Is there reason to think that such a trained eye could also be used to objective conclude if police officers engaged in appropriate or excessive violence? Is that in turn something that can be taught to a model?

@ihsiehchi
Copy link

ihsiehchi commented May 29, 2020

The killing of George Floyd

First off, my apologies for making the final week's discussion gloomier than it should be, but for me, it is impossible to read a book chapter about violence without constantly making connections to the source of chaotic unrest for the past two days and for days to come.

My question is a somewhat applied one without the specific setting of police brutality. Some have asked the questions, "why didn't the other three offices intervene when the fourth officer had his knee on Floyd's neck?" The book provides a reductionist, generalized by nonetheless what I think is an adequate answer when introducing the concept of "emotional energy":

These are positive experiences when all participants feel solidarity
and intersubjectivity. In these successful interaction rituals, individuals
come away with feelings of strength, confidence, and enthusiasm for
whatever the group was doing: these feelings are what I call emotional
energy.

If I want to test whether this is indeed the case, as a first step I would focus my analysis on the facial expressions of the officers in the frames around the beginning of any aggressive behavior. A challenge may be that one may express highly similar facial expressions in fear and in enthusiasm. I wonder whether machine learning methods would be capable of distinguishing those facial expressions since it requires the model to be trained on a large number of photos of the officers. Furthermore, not only is the task virtually impossible for human beings, any attempt at the task, I imagine, would be indubitably subject to substantial biases due to the nature of the event under investigation.

@jsgenan
Copy link

jsgenan commented May 29, 2020

I am still trying to digest "violence is a situational process". Nevertheless, I guess it is secure to say the goal of the read is to objectively record and interpret a violence incidence. The photos and videos might be more competent to the task, but I want to comment that all cameras are emotional too. Otherwise we won't see any value in journalism photography and documentary filming. The timing, light, angle of filming matter a lot to interpretation.

@Lesopil
Copy link

Lesopil commented May 29, 2020

My question continues the thoughts of several previous commenters. I think that it is broadly true that a photo or video can record/preserve the event better than a verbal/written recollection, however even in a photo there is bias. There is the famous photo from the Vietnam war of the south Vietnamese general executing a viet cong prisoner. Depending on ones political alliance, the photo could be interpreted differently. There is a lot more interpretation that has to occur by human agents when examining photos, as opposed to text. I wonder, looking back to last week, if it is possible to do something similar. Say, predicting the next frame of a movie, or something like that?

@liu431
Copy link

liu431 commented May 29, 2020

My question is what are researchers' intent to study violence? If the focus is on understanding the root cause of violence, how could it be more helpful for influencing policymaking?

@minminfly68
Copy link

This paper is super interesting and thought-provoking that provides me with a new understanding of violence. I partially agree with author's points about how violence develops and its intention. I just hope author might can explain more about the mechanism of violence as well as the intersection between violent and non-violent cases like whether we could expand this model to a more general case study? Thanks!

@bazirou
Copy link

bazirou commented May 29, 2020

The paper introduces the meaning and empirical experiment results of dimension in culture. My question is, how can we interpret the meaning of dimension in culture?

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

No branches or pull requests