Illustrating geons

My first year as a cognitive science student presented me with a fascinating, if not overwhelming overview of the inner workings of the mind, but one of the concepts that stood out the most was geons. A geon is a "geometric primitive," or one of several geometric shapes that can comprise any object. Humans then recognize and combine geons to identify objects: once a group of geons is recognized as an object -- say, a sphere and a triangle for an ice cream cone -- it is stored in memory. New input will be matched against stored representations during object recognition. The existence of geons has not been proven (though there is considerable evidence for it), but is part of one of several theories of how people interpret and process visual information. In this particular case, recognition-by-viewpoint theory postulates that visual input maps to representations of imagery, such as geons, in the brainI thought it incredible that our interpretation of complex objects could be reduced to a few geometric primitives –– a circle, a square, a cylinder.

I first learned about geons in a class on perception and visual arts, a tri-weekly adventure in optical illusions, neurons, and perspective in Western art. We were in the midst of discussing various ways to interpret visual information: what if something looks different from each angle? What if it looks the same? How do we contend with an incredible amount of information in our memory? The idea is that geons are part of a relatively extensible process -- we can resolve components into an infinite number of objects, making it easier to learn, categorize, and identify any visual input around us.

This clock can be resolved from a handful of geons.

This clock can be resolved from a handful of geons.

For instance, speaking in terms of geons, this clock is a large circle and two rectangles. When you see the clock, your brain recognizes the geons and then resolves them into the clock. Remarkably, is it thought that there are about 36 geons, each highly distinct. 



I think about geons a lot when I design. If I have a lot of visual information, how does my brain parse and select between it? What kinds of visual cues can I provide to give a better experience?

Flat design is interesting because it abstracts complex objects into lightweight, often symbolic components. If there is no need for intricate visual detail then leave it off. In iOS 8, many icons are small geometric units combined. We're probably seeing geons anyway; why not make the whole world look like them?.

A few weeks ago I decided to draw bicycles in Illustrator. At first I was overwhelmed; bikes are enormous and extremely complex. There is the twisting frame, chains, the seat, lots of metalwork, and even accessories to account for. I thought of reaching for a graphics tablet but eventually, in a moment of clarity, decided it would be simpler and more direct to use geometric shapes. Interestingly enough, breaking the bike into small shapes made me appreciate (and maybe understand) its composite construction more. Bikes have spokes, handlebars, gears... what does each component look like? Even though my bike illustrations were fairly simple, I was forced to think about each part, each shape, and how it related to the others. And as the bikes came together, I wondered whether what was happening on the screen in front of me -- little geometric shapes combining to make a complex object -- paralleled what happened in my brain when I went out in the world and looked at the bicycles around me. If geons show us anything, it's that object recognition, like the process of creating bicycles, is at once enormously complicated and incredibly simple.

Usability and Mobile Transit Apps

"Smart Ride," "YourBus MBTA," and have collectively saved me quite a bit of time. Many days, I take a bus to and from work, and due to the unpredictable circumstances of this particular route, its published schedule is seldom accurate. My mornings became a little less harried with the discovery of a suite of transit tracking apps.

There are two flavors of transit apps: the Boston-specific apps (, YourBus MBTA), and the all-purpose apps (Smart Ride, NextBus). At first I loved the extensibility of Smart Ride and others of its ilk -- on several trips to San Francisco, I was able to pull out my phone and navigate the bus system while far out in the Sunset. The app was the same, but the geography had refreshed. It was a pleasant surprise not to have to download a new app for a new city.

From the Smart Ride website.

From the Smart Ride website.

But upon my return to Boston, I noticed something peculiar in Smart Ride: the tracking system had become so sensitive that it had begun to include the Harvard and MIT shuttle schedules. These shuttles are for those affiliated with the Harvard and MIT community, and for those users, it may be very useful to have public and institution-specific  transit information integrated into one app. However, for the rest of the community it was noise. Moreover, there was no option to remove other forms of the transit. NextBus has this down a little better -- its content refreshes depending on where you are, but you still choose the route.  Both apps show all the agencies supported, which seems more of a testament to their growth and is less immediately useful to the user. It seems that both of these apps overwhelm the user with choice, compromising findability. There are so many options that it can be hard to identify the one they are seeking.

Smart Ride has a certain level of customizability: there is a "Favorites" tab where users can add frequently-used routes. Nevertheless, it's useful largely for a regular commute and not when you find yourself across town and just need to get somewhere.

So the question remains: how much content is too much content?

Last summer I held a UX design workshop in conjunction with Ladies That UX Boston. A city of trains, bicycles, and buses, our focus was naturally public transit. We talked about which transit apps worked for us, and which didn't. We discussed having not enough choice: using a bus-only app when you also take the train sometimes, versus too much choice (a hyperactive Google Maps that shows you many possible routes with all modes of transportation when all you want to do is go to work in the morning).

"I just want to know what's fastest," someone lamented. "But I also want to take the bus or train because I don't have a car." She paused. "Or maybe bike, I guess. If it's nice out. Maybe the app could tell me the weather."

We spent the rest of the evening brainstorming and sketching, and I concluded that the user experience is should be easily customizable. My transit app should show me buses and trains tailored to my geography, and just that. After all, the app should appeal to the average user, who is probably not a student at a particular university (not to mention that there are over 250 colleges and universities in the Boston area!). But let's offer options. Let students toggle on and off their campus' shuttle, and make that option discoverable, ensuring that the functionality is easy to find and its value is readily understood. 

One final thought: one of the key values of city-specific apps is that they can provide language, directions, or traffic warnings that are meaningful to residents of that city. Many GPS services are not sophisticated enough to understand how strange Boston's layout is –– what appears to be the shortest path is often not, and many landmarks go by local names (the "Harvard Bridge" is known as the "Mass Ave Bridge," which describes its location more accurately).

For now, I'll keep using Smart Ride because it serves most of my needs locally and when I'm out of town. But I'd advocate for a more customizable interface that focuses on what's here and now, rather than what might work in another city or if the user has access to a semi-private form of transportation.