Introduction
“Nothing is more humbling than to look with a strong magnifying glass at an insect so tiny that the naked eye sees only the barest speck and to discover that nevertheless it is sculpted and articulated and striped with the same care and imagination as a zebra. Apparently it does not occur to nature whether or not a creature is within our range of vision, and the suspicion arises that even the zebra was not designed for our benefit.” ― Rudolf Arnheim
We have all heard the expression, “The eyes are the window to the soul”. While the context of that expression may be up for some debate, the fact remains that our eyes are certainly a gateway to the mind: how we ultimately perceive what we see, and how our perceptions remain cemented in consciousness.
The eyes are incredibly powerful, sensory organs that process a continuous flood of information, all day, every day. And how they handle that information is aptly called visual information processing. As a short side bar, the auditory processing system – the interaction between ears and the brain – can also contribute to images formed within the visual cortex, located in the occipital lobe (Revlin, 2013).
For the purposes of this paper, though, we will focus on how visual information processing works through the eyes and then into key regions of brain. In addition, we will look at how bottom-up and top-down processing handle visual stimuli, how conditions such as hemineglect and prosopagnosia impede visual processing and the relationship between attention and perception, and how current research may help further our understanding of visual information processing.
A Short Overview of Visual Information Processing
Before we dive into bottom-up and top-down processing, we must first understand which triggers help begin the process. First, it is well-established that we humans are a highly pattern-seeking species. Like other mechanisms that drive our anatomical and biochemical makeup, our need to seek patterns in nature, and within our own cognition, is rooted in evolution: Pattern recognition is not only a method for perceiving the world around us, but it is also a survival tool for identifying potentially dangerous situations and predators (Revlin, 2013).
Our senses are the initial gateway for visual processing: For example, you are at work, diligently managing project you need to complete by the end of the day, when, in your periphery view, you see your manager approach you. You immediately turn your attention to your manager, who waves down two of your co-workers to walk over and join the conversation about to be had.
In other words, all within in a matter of seconds, you moved your attention – like a spotlight shifting its focus — from your project to your manager standing now standing in front of you. Your eyes, the key sensory organ in this case, feed all pattern details into your visual cortex (Revlin, 2013).
And because the visual cortex must process a large amount of information all at once, it uses a combo streaming/buffering system, called sensory storage, to help manage the incoming flood of sensory input. (Revlin, 2013). In effect, sensory storage is like a security guard directing traffic at a concert arena: as people stream into the area, the guards ensure that people do not storm the arena or cause other kinds of confusion. Therefore, the guards buffer the traffic.
This also allows the brain to use two key methods for gathering visual stimuli once they are captured by the senses: bottom-up processing and top-down processing. In short, bottom-up processing and top-down work together to organize visual input into a hierarchal format, as well as ensure that both processing methods are working in harmony (Rauss & Pourtois,2013).
Now, let’s explore bottom-up and top-down processing in more detail:
Bottom-Up Processing
Once the eyes are engaged, and attention suddenly focuses on a specific person, place, or thing – or all of the above! – bottom-up processing starts with gathering basic details. To illustrate this, let’s go back to the example of you and your manager at the office: When he first approaches your desk, you notice that he is wearing a suit, and he is smiling at you as he waves at your co-workers to stop over and chat. These details, of course, are rather rudimentary. However, bottom-up processing is just that: it manages the basic details, which are then fleshed out with top-down processing.
Yet bottom-up processing should not be oversimplified. Even though it concentrates on visual stimuli basics, its function relies on four important sub-processes: distinctive features, recognition by components, template matching, and prototype theories (Revlin, 2013). Let’s explore each sub-process.
Distinctive Features
Because we are pattern-seeking creatures, a way that we determine visual patters is called distinctive features. That is, all stimuli are distinct from each other; these distinctions are known as features. The distinctive features process initially homes in on more important features, and then compares/contrasts them with memory stores in the brain. The compare/contrast process often involves taking more familiar, easier to identify stimuli, called global features, and then using that information to interpret the finer details, called local features (Revlin, 2013). We could even say that the global and local features relationship is its own bottom-up / top-down method within the main bottom-up/top-down mechanism.
Recognition by Components
As we live in a three-dimensional (3-D) world, our visual processing system helps us view the world in 3-D. Recognition by components bridges the gap between the brain and 3-D objects. In turn, the visual processing system breaks down the objects by organizing them into sub-components (Revlin, 2013).
For example, a football has a particular 3-D shape, with lines and thread defining the overall shape. This essentially allows the visual processing system to construct the components into their full 3-D form.
Template Matching
Like other systems in the brain, memory stores are powerfully dynamic and constantly retain an extraordinary amount of information. By using “templates” from other experiences that may match up with the visual stimuli we receive at any given moment, we can draw on our memory stores to decide whether we do in fact recognize the stimuli, or that our memory stores need to generate a new template to be used for future experiences (Revlin, 2013).
For example, if you see your co-worker at the gym on a Saturday afternoon, you use a storied template of your coworker’s features to help conclude that the person at the gym is one in the same.
Prototype
Although template matching, along with other bottom-up processing functions, has its value, it also has its limitations. What if, say, you had mistaken the person you saw at the gym for being your co-worker? Let’s say that both your co-worker and the stranger at the gym have the same hair style, similar facial structure, and a strikingly similar body type: The template that bottom-up processing employed was clearly faulty.
However, because physical features often overlap among people – similar noses, similar eyes, similar height, etc. – bottom-up processing also uses prototypes to classify features into broader categories (Revlin, 2013): other people have a similar hair style, other people have big green eyes, other people may have similar proportions.
From there, prototypes can be further defined, yet still broad to keep the context accurate. For instance, a domestic tabby cat is part of the greater cat family, but it is not a lion. Yet because the prototype for both is so remarkably similar, we can call both a cat.
Top-Down Processing
We can see that, for dealing with the more basic end of visual input, bottom-up processing has its own set of detailed functions. The same functions aggregate up to top-down processing – which helps fill in the remaining details by more precisely analyzing them and making the visual input more complete. We can especially see this in action when speech is involved, as well what is called the word superiority effect. Let’s look at each example.
Speech
Because speech involves processing sounds. It also gives us insight on how we start with simple information and then process it into a more complete understanding (Revlin, 2013). For example, your friend calls you up one day and begins the conversation with, “Hi, what are you up to?” Your brain not only processes the voice you recognize, but also the words themselves, which are familiar to you.
If English is your first language, a simple phrase such as “hi, what are you up to” is an easy given for bottom-up processing, Then, top-down processing is engaged to analyze the inflection in your friend’s voice, whether it is pleasant or perhaps impersonal, and what Is the overall context of your friend asking the question. Or your friend, may use a slang such as “whaddup?” – which also requires top-down processing to determine that “whaddup” equals “what are you up to”.
The Word Superiority Effect
In fact, that example of “whaddup” segues into what is called the word superiority effect. Namely, we are able to recognize patterns and words and then process them accordingly. This is even more evident when a string of words is missing letters or is grouped in an unconventional way. Example: when you first read the sentence, “Mary had a little lamb”, if the sentence is written instead as “A little lamb Mary had”, or “Mary hxd a lttle lamb”, top-down processing would help decipher that the sentence context is “Mary had a little lamb”.
Examples of Disorders that Affect Visual Processing
Unfortunately, the human brain is prone to error. Whether it be from physical damage / trauma, a developmental disorder, or a genetic factor that causes neurological dysfunction, the brain must find ways to compensate for its deficits. Let’s explore two disorders – hemineglect and prosopagnosia – that interfere with visual information processing and overall brain function.
Hemineglect
The brain is divided into two main hemispheres: right and left. Each hemisphere controls opposite sides of the body: the right hemisphere controls the left; the left hemisphere controls the right. In addition, the parietal lobe, which occupies space parts of the left and right hemispheres, helps manage visual input as well as language functions (Ogden, 2012).
However, the left parietal lobe and right parietal lobe differ somewhat in function: Because the left parietal lobe deals heavily with language, it primarily leans on the right parietal lobe to help manage both the left and the right. Therefore, if the right parietal lobe is damaged, the left visual field also becomes damaged or absent altogether (Ogden, 2012). This is referred to as hemineglect.
Interestingly, though, hemineglect is more of a problem with attention and memory than a full-on visual impairment: Hemineglect can spawn from direct trauma to the parietal lobe, or from a certain dysfunction that develops over time. Either way, hemineglect yields many of the same results. Here are two examples:
You have hemineglect and you are sitting at a table: Your cell phone is moved from your right field of vision to your left; therefore, you may no longer be able to see the phone. However, if you are asked where your phone was located before it was moved, your initial attention to the phone, as well as your memory stories, prompts you to answer that your phone was originally on the right side of the table.
Or let’s say your right field of vision is directed to a couple of books sitting on the left side of the table: Your right field of vision is then moved back to its original focal point on the table; and when you are asked to recall what is on the left side of the table, you most likely will answer that a couple of books are sitting on the left end of the table.
Yet hemineglect can still be a rather debilitating dysfunction. In extreme cases, a person with hemineglect may even have trouble identifying with any physical features on the left side of the body (Revlin, 2013).
Prosopagnosia
As we know that hemineglect can disrupt attention and perception, what if the condition involves not being able to recognize familiar faces? This condition is called prosopagnosia – which affects approximately 1 in 50 people (Psychology Today, 2020). And like hemineglect, prosopagnosia does not necessarily stem from inherent issues with visual and memory processing (Revlin, 2013). Rather, it typically stems from damage to the fusiform gyrus, located in the temporal and occipital lobes.
It is particularly striking that although people afflicted with prosopagnosia have trouble recognizing other people by their faces, they can more easily identify voices and basic facial features. And they can even determine whether a person has younger or more elderly features (Revlin, 2013). Yet they still may not fully recognize the person standing in front of them.
One explanation for this is that the visual processing system is not well-developed in capturing or managing visual prototypes created during bottom-up processing. Therefore, those afflicted with prosopagnosia have trouble applying broad categories of features to more detailed ones (Revlin, 2013)
Current Trends in Visual Information Processing Research
While advances in imaging technology (fMRI, EEG) have helped us gain a more thorough understanding of sensory information processing and sources of neurological disorders, the research is not close to complete. There is still much to learn.
For example, recent studies regarding the relationship between top-down and bottom-up processing have shown that both processing mechanisms work much more as partners than in silos. The key to this interactive relationship is a component called predictive coding, which basically acts as a facilitator between bottom-up and top-down. Research is still being done to validate this further – in the meantime, it is now more clearly established that bottom-up and top-down are not opposites, and that together they complete the hierarchies found in pattern recognition (Rauss & Pourtois, 2013)
Progress has also been made in diagnosing and treating neurological disorders that affect visual processing. For instance, prosopagnosia can sometimes be a bit tricky to diagnose; basic screening can be a bit faulty. However, a detailed fMRI can help pinpoint the results. In addition, a rather recent test, called the Cambridge Face Memory Test, has proven to be more thorough than basic screening tests. Yet like other tests, it does not have a 100% success rate in diagnosing prosopagnosia. Still, having multi-varied options on-hand can provide a better means to diagnosing the issue (Psychology Today, 2020).
Conclusion
Visual Information Processing is a complex system that involves two main processing systems: bottom-up and top-down. Together, bottom-up and top-down processing bridge our attention focus with what we ultimately perceive. In short, bottom-up and top-down allow us to form complete pictures of visual stimuli.
For some people, processing visual information can be more challenging. Neurological disorders are often at the core of the issue, though they may not always involve direct damage to the visual cortex. Instead, damage to a particular region of the brain can impede visual processing, but not eliminate it. Prosopagnosia, for example, is due to a damage to the fusiform gyrus, which specifically manages facial recognition. Hemineglect is typically due to damage to the right parietal lobe.
Overall, whether a neurological disorder is involved or is not, we know that the brain relies on several regions to partner up when processing sensory input. And it is just as clear that our eyes are a primary driver in information processing.
References
Adaval, R., Saluja, K., Jang, Y. (2018). Seeing and thinking in pictures: A review of visual information processing. https://onlinelibrary.wiley.com/doi/full/10.1002/arcp.1049
Ebaid, D. (2019). Visual Information Processing in Young and Older Adults. https://www.frontiersin.org/articles/10.3389/fnagi.2019.00116/full
Revlin, R. (2013). Cognition theory and practice (1st ed.). New York, NY: Worth Publishers
Rauss, K., Pourtois, G. (2013). What is bottom-up and what is top-down in predictive coding?. https://www.frontiersin.org/articles/10.3389/fpsyg.2013.00276/full
Ogden, J. (2012). The Bizarre Disorder of Hemineglect. https://www.psychologytoday.com/us/blog/trouble-in-mind/201202/the-bizarre-disorder-hemineglect
PsychologyToday. (2020). Prosopagnosia. https://www.psychologytoday.com/us/basics/prosopagnosia