If you're seeing this message, it means we're having trouble loading external resources on our website.

If you're behind a web filter, please make sure that the domains *.kastatic.org and *.kasandbox.org are unblocked.

Main content

Bottom-up vs. top-down processing

Bottom-up and top-down processing are two different ways of making sense of stimuli. In bottom-up processing, we allow the stimulus itself to shape our perception, without any preconceived ideas. In top-down processing, we use our background knowledge and expectations to interpret what we see. Created by Ronald Sahyouni.

Want to join the conversation?

  • leaf green style avatar for user Sagar Shah
    Where's waldo is a poor example describing this concept
    (3 votes)
    Default Khan Academy avatar avatar for user
  • piceratops ultimate style avatar for user ArminBay
    For all of you who had a problem understanding this, I'll simplify this video in two sentences:
    1. Bottom-up processing is when the environment (stimuli) influence our thinking
    2. Top-down processing is when our thinking influences how we see (understand/perceive) the environment.

    Source: Psych Undergrad
    (40 votes)
    Default Khan Academy avatar avatar for user
  • leaf green style avatar for user Brooke
    Is the presence of a goal all that is needed for something to be considered top-down processing? For example, I know that the goal of the cockpit (at ) is to fly the plane. But despite knowing that purpose, I still have no idea what I am looking at. Everything is new and novel, even if I am aware of the goal. In the case of where's waldo though, I know the goal is to find waldo, but that knowledge doesn't actually help me find him faster. I still need to go through the page piece by piece (which seems to imply that it is bottom-up processing) in order to find him. It seems like a top-down processing strategy for where's waldo would be using some prior knowledge or experience of where waldo is typically hidden to find him, or otherwise using some kind of strategy to locate him. In other news, I had absolutely no idea that there were strategies for finding waldo until I googled it just now. I can't decide if it is brilliant or cheating.
    (26 votes)
    Default Khan Academy avatar avatar for user
    • That's a very good question. To be honest, I try not to see top-down processing as merely goal-driven but rather that it is a type of processing that relies on prior knowledge and experience. Top-down and bottom-up processing are processes that usually work together, so it's not normally this or that happening at one time. You may be using top-down processing as you perceive an object whose form you're familiar with (like the cockpit), but at the same time, also be using bottom-up processing to make sense of unfamiliar things pertaining to the object (like trying to determine what some buttons on the cockpit do based on their observed properties or position or whatever).
      (22 votes)
  • leaf green style avatar for user Sydney Vogt
    The examples given in this video make the bottom up vs. top down processing very confusing. I supplemented with another resource and from what I am understanding I believe a better example of top down processing would be that you know someone who is usually very mean and they make a joke, because you know how they are, you will assume that they said the joke to be mean rather than funny. This is the best example I can find of that uses prior knowledge (in this case the prior knowledge of the mean person) when processing the information. Is this a correct way of thinking about top down processing?
    (16 votes)
    Default Khan Academy avatar avatar for user
    • leafers tree style avatar for user Ryan
      Correct, the scenario you described is an example of top-down processing. An example of bottom-up processing would be if you had met that same person (but WITHOUT the pre-conceived notion that they are mean) and assumed their joke was funny based on their presentation of the joke.
      (13 votes)
  • marcimus pink style avatar for user Mattie
    On one of my mcat practice questions it talked about drivers connecting traffic lines as they drive across a street. Their answer was Gestalt Principles because the driver had to make a connection out of nothing, however this seemed to resemble the cube example of top down processing. What is the key difference?
    (4 votes)
    Default Khan Academy avatar avatar for user
  • mr pink red style avatar for user Turk.Sad
    Per bottom-up processing, this stimuli in this video is confusing me :-/
    (4 votes)
    Default Khan Academy avatar avatar for user
  • winston default style avatar for user adinsx0001
    Is the dress black and blue, or is it white and gold? Is top down processing responsible for why some people see the dress one way and not the other?
    (3 votes)
    Default Khan Academy avatar avatar for user
    • leafers tree style avatar for user Ryan
      Maybe it's a combination of some of the other concepts we've learned. It makes sense that top-down processing would result in the different perceptions of the dress, but perhaps people have different thresholds for perceiving certain colors as well.
      (2 votes)
  • blobby green style avatar for user Jackie Wooley
    Is this video really trying to get at the Epiricist/Idealist philosophies as they apply to neuroscience? From what I can tell, "bottom up" processing is more of an empiricist perspective (Locke, Hume) and the "top down" is more of an idealist perspective (Kant, Descartes).
    (0 votes)
    Default Khan Academy avatar avatar for user
    • blobby green style avatar for user Navid Darayan
      Think of it likes this:
      Bottom-up processing: begins with the sensory receptors and works up to the brain where the stimuli is integrated and processed. (So something new to you)
      Top-down processing: Brain applies experience and expectations to interpret sensory information. (So something that you are familiar with).
      (7 votes)
  • blobby green style avatar for user tcoyle
    Adopting a more philosophical perspective, it seems likely that there is no such thing is "pure sense data". You can't approach something with out preconceived concepts. I'm not sure this epistemic distinction between BUK and TDK holds.
    (2 votes)
    Default Khan Academy avatar avatar for user
  • blobby green style avatar for user Kavya
    Which processing is better? Bottom up or top down?
    (1 vote)
    Default Khan Academy avatar avatar for user

Video transcript

Voiceover: Let's look at the difference between top-down and bottom-up processing. So, what is bottom-up processing? Bottom-up processing basically begins with the stimulus, so let's imagine that we're looking at something, or let's say I'm looking at a banana. The banana is sitting there and it influences what we perceive. So, stimulus influences what we perceive, our perception. So, if I know absolutely nothing about something, then the stimulus or whatever it is I'm looking at, yet I don't know anything about, I've never seen it, I don't have any preconceived cognitive constructs about what it is I'm looking at. The stimulus basically is influencing my perception. So, for example let's imagine that I'm looking at a cockpit of a plane. I'm not a pilot, so I'm not really too familiar with everything and everything kinda looks fairly confusing. So, basically all the different stimuli, so this stimulus, a bunch of gauges, and this rudder-looking thing, I'm basically looking at all the different little parts of something that is new and novel to me, and trying to kind of comprehend what it is I'm looking at. So, this is bottom-up. This is when you start with no preconceived idea of what it is that you're looking at, and allow the stimulus to influence your perception of what it is that you're looking at. So, bottom-up processing is data-driven, and your perception of what it is that you're looking at directs your cognitive awareness of the object. So, in contrast, top-down processing basically uses your background knowledge, so uses your background knowledge to influence perception. So, let's look at this example over here. So, what we're actually seeing are a bunch of circles, they are just a bunch of circles and then inside the circle there are a couple of lines drawn. So, we are looking at this set of circles, these white circles with lines drawn inside of them. We are creating this cube. We're basically taking these lines and then putting them together in order to create a cube. Even though the stimulus itself, which is the circles with the lines, actually doesn't draw a cube because there are these black spaces over here, and there is absolutely nothing in the black spaces, but our brains are basically taking this information and using our knowledge of cubes and what they're supposed to look like, we're recreating a cube despite a lack of a cube actually being present in the image. So, that's top-down processing. It's using your background information, your background knowledge, your learning, your expectations, in order to influence what it is that we're perceiving. So, in other words, it's theory-driven. We look at this and we assume that they're trying to represent a cube, even though one's not actually drawn there, and we're using that theory in order to shape our cognitive understanding of what it is that we're looking at. So, our perception, our behavior is influenced by our expectations, which is top-down processing. So, we're using what's already in our heads in order to perceive what it is that we're looking at, whereas in bottom-up processing we're using the stimulus itself in order to drive our perception. So, another good example of top-down processing would be "Where's Waldo?" So, in "Where's Waldo?", we have a mental idea of what we're trying to do, which is to find Waldo amidst this really jumbled mess of a picture. So, if we were using bottom-up processing in order to look at this we would just be seeing a whole bunch of little people, and we wouldn't really be goal-driven, we wouldn't be trying to do anything, but with top-down processing we have a goal, and we're able to look through here to find Waldo.