Hello Fancy Folks! Welcome back to LuxeSci, a podcast to re-ignite your wonder by exploring the intersection of science and luxury. I know it’s been a little bit since we’ve released a new episode. We’ve taken the time off to not only re-think this season, but also just to get into new routines, meet new people and host a ton of holidays.
But it’s a new year and I won’t say it’s a new us. We’re still here, we’re still super curious about everything around us and we’re still fascinated by the finer things in life. As promised, Season 3 is all about the science of fashion. There are so many different ways science influences and is embedded in the processes of the fashion industry that it was challenging to decide which ones to focus on and to develop a season that felt cohesive (shout out to my Project Runway fans!).
I decided to start with one of the less tangible elements of fashion. It makes sense that science would underpin cloth creation, dying, construction, etc but what about why we choose the clothes that we do? What does science have to do with fashion trends? (I’d like to thank Sheeva from Fancy Comma for the suggestion).
Background
Defining fashion is a bit of an undertaking. The word has grown to mean anything an industry, a style, an aesthetic and a trend. It describes the creation of clothing, footwear, accessories, etc and their mix and match into outfits
The term fashion is from the Latin “Facere” meaning to make
Mode (french) dates back to 1482 while the English version meaning “in style” dates only to the 16th century
Fashion can be used to differentiate (new fashions) or to be inclusive (the fashion of the 1950s)
Describes the social and temporal system that influences dress as a social signifier in a certain time and context
Style - expression that lasts over many seasons and is often connect to cultural movements and social markers
Trend - group of products or a style that becomes fashionable for a period of time and influences masses of customers
Not going to go into the history of fashion. There are some great books and articles about that very topic. And as much as I like history, we’re not here for that, we’re here for the science.
Science
So I hope the definitions gave you a little clue as to the topic of this episode. We’re going to be focusing on the science of trends. How to predict them and why they occur.
First some basic orientation around trends. There are different types of trends:
Fads - usually last 3-6 months (micro trends)
Real trends - 6 months to 5 years (macro trends)
Seasonal 6-12 months
Long-term 1-5 years (usually focus on particular key articles whose form changes slightly over the seasons) e.g. skinny jeans
Classics - 5-25 years (e.g. little black dress)
There are several models of how trends evolve. One such model is called the Diffusion of Innovation model by Everett Rogers
Starts with an individual exploring a new way of dressing
Then a social group accepts this new way
It is picked up by designers and makes onto the catwalks of fashion shows
From there it goes to brick-and-mortar and online stores
Then it’s all over the streets
The phases of the model are called
Innovators
Early adopters
Early majority
Late majority
Laggards
So why trends? What is it about humans that produces trends?
It has a lot to do with social psychology
Branch of psychology that deals with social interactions, including their origins and their effects on the individual
2 social psychologists (Henri Tajfel and John Turner) developed a social identity theory in 1986 that includes in-group and out-group
Individuals experience collective identify based on their membership in a group
Leads to the categorization of “us” versus “them”. In order to maintain a positive social identity, people will engage in intergroup comparisons that have a favorable bias to their in-group
This self-categorization can be so powerful that it can be activated automatically with subtle stimuli (i.e. what someone is wearing)
Applying this to fashion, the in-group can be those following a trend or those not following a trend depending on your point-of-view.
The tendency for individuals to follow trends set by celebrities or people they look up to can be accounted for by tendency for people in in-groups to bolster themselves by following people in in-groups that they have a positive affinity for.
The concept of in-group and out-group and the behaviors ensuing from these distinctions can feel very negative, especially in today’s polarized environment and especially for people in minority groups. As with much of human behavior, it isn’t all bad. In and out groups can create a sense of unity or harmony with others, a sense of belonging.
Another psychological reason we are inclined to follow trends is simply that its easier. Our brain is a powerful machine and as such, looks for shortcuts to make things easier. By following others, this shortens and eases thought processing.
Instead of having to decide what to wear, it’s easier to have someone else decide for us (honestly, I would definitely have a stylist if I could afford one).
One big pull for trends is their novelty. Jennifer Baumgartner sums this up nicely in a piece for Psychology Today. When we see something new, the reward center of our brain (the substantia negra/ventral tegmental area, which is the area that produces dopamine. It also controls your ability to learn and think)
A study done in 2006 by Nixo Bunzeck and Emrah Duzel showed that we seek novelty to receive a reward. And, given that the part of the brain that controls reward also controls learning, the study also showed that learning is enhanced in the context of novelty.
So its natural that we would gravitate to something new and a new trend in fashion. Not too sure what the learning aspect has to do here but that would be interesting to follow-up on
So now that we understand a bit of why trends happen, at least psychologically, how does the fashion industry forecast trends.
Historically, fashion trends have been predicted on the macro level and were largely driven by style icon. Fashion houses would analyze what these individuals were wearing and trends tended to last longer
In the 2010s, the burgeoning field of data analytics was applied to various marketing endeavors, including fashion. Companies could look at demographic specific reactions to past products and target age groups with marketing strategies.
Used 3 main types of analysis for this
Descriptive analysis - summary of past data to observer previous trends
Predictive analysis - use historical data to predict future trends
Prescriptive analysis - analysis to help guide future decisions
Really great example of data analytics for fashion trends in the Illum Magazine from USC Viterbi school of Engineering
These are powerful tools for both fashion houses and fast-fashion providers. So how could it work?
Researchers from Seoul National University used network analysis to probe fashion trends from 2004 to 2013 and published the results in the Journal of the Korean Society of Clothing and Textiles in 2014
Network analysis - looks at relationships among entities. It can describe and make inferences about relationships of individual entities, subsets of entities and of entire networks
In this paper, the entities were individual fashion trends and a 2-mode network was created consisting of seasons and trend languages derived from the Samsung Design Net.
What they found was interesting
From 2004 - 2008, retro modern, feminine modern and ecological modern trends were dominant
Starting in 2009, the trends that were popular were around natural style
They saw a re-emergence of the summer/spring trends every 2 years that allowed for a prediction that S/S 2014 would be another year for a version of natural style
Over those 10 years, the macro trend was modern and natural, which I think is the direct predecessor of today’s quiet luxury looks
So if data analysis is taking over the fashion world, what about Artificial Intelligence (AI?)
You can use AI models to scan tons of images of runways and social media to pull out the most popular styles, colors, etc
Lai and Westland published such a model back in 2020 that used supervised and unsupervised machine learning to pick color palettes for runway shows
This method showed close agreement with the data from 22 participants asked to pick 3 colors to match each of the 48 images from a runway show
Supervised vs unsupervised machine learning has to do with the data inputted into the model for the AI to train on
Supervised learning uses labeled datasets. This allows the model to measure its accuracy and learn over time
Unsupervised learning uses unlabeled data - this is used to analyze and cluster unlabeled datasets. Discover hidden patterns in datasets without human intervention (labeling)
So why does this all matter. Is this just another case of someone with a hammer thinking fashion images are a nail? With the fashion industry throwing out more than 92 million tonnes of fabric each year and potentially accounting for 4% of global emissions, more accurate prediction of trends could result in meaningful reductions in waste, especially since our appetite for fast, cheap fashion doesn’t seem to be waning.
Glossary
Trend - group of products or a style that becomes fashionable for a period of time and influences masses of customers
Supervised machine learning - when the data analysis model uses labeled data to be able to check its accuracy
Unsupervised machine learning - when the data analysis model uses unlabeled data to discover new patterns in the data
Fun Cocktail Party Facts
The fashion industry throws out more than 92 million tons of fabric annually but AI model predictions of trends can help decrease that
Our love of trends has to do with social identity theory of in-groups and out-groups and how novelty sparks learning in our brain
I hope you enjoyed learning about the science that underpins fashion trends. I know I’ll think a little bit more about why I want to buy that new Coach bag that’s all over my Instagram feed. As always a special thanks to my audio engineer, Dr. Dimos. Our theme music is Harlequin Mood by Burdy.
And thank you for listening. If you like us, please share the podcast with friends, family, co-workers, etc and find us all over social media at Luxescipod. While you’re there, drop us a line and say hello. We love hearing from you.
References
Comments