Cognition – Chapter 3
- Human vision is the dominant sense; more brain area is devoted to vision. We tend to
trust our vision over other senses when our sensory input conflicts.
- Form perception: the process through which you manage to see the basic shape/size of
- Object recognition: the process through which you identify what the object is.
Why is Object Recogntion Crucial?
- Crucial for applying your knowledge to the world (& so to take on an action based on
what you know about shoes, phones, etc.)
- Also crucial for learning – in order to combine information collected on different
occasions about a person or a thing, you must be able to identify it.
Beyond the Information Given
- Gesalt psychologists: organization must be contributed by the perceiver, which is why
the perceptual whole is often different than the sum of its parts.
- “beyond the information given” – ways that our rperception of a stimulus differs from
(& goes beyond) the stimulus itself.
- Necker Cube: reversible figure – can be percieved in 2 different ways. The lines are the
page are neutral with regard to the shape’s configuartion in depth (no ‘proper’
interpretation). Your perception is not neutral though, therefore it going beyond the
information given in the drawing by specifying an arrangement in depth.
- The vase/mirrored face picture is neutral with regard to figure/ground organization – the
determination of what is the figure (object) & what with the ground (background). Again,
however, your perception is not neutral – your perception contains information (about the
arrangment of depth or which part is figure or ground) that is not contained in the
stimulus itself. Therefore then, this is information contributed by you, the perciever.
- One may argue that the reversible figures are special – carefully designed to support
However this is wrong, because many stimulu are ambiguous & in need of
interpretation. We don’t often detect this ambiguity because the interpretation is done
- Ex. bowl of fruit; assuming different fruits extend/connect in places you can’t
actually see – this is how you interpret them, it is not guaranteed.
- However, your interpretation is not careless, it is guided by a few straightforward
principles (catalouged by Gesalts).
- Perception is guided by principles of proximity & similarity. If you see elements that are
close to eachother or elements that resemble each other in the same visual scene, you
assume these elements are parts of the same object. You also tend to assume that contours
are smooth. - Everyone’s perceptions are guided by the same principles, which is why you generally
perceive the world the same way as everyone else.
Organization & “Features”
- Interpretation of the input sometimes happens before we start cataloguing the basic
features of an image.
Ex. interpret black marks as spelling the word “perception” because you “provide” the
missing lines in the image. Once you’ve interpreted these absent lines, you can then “fill
in” the missing features.
- With one organization, the features of an image may be absent, but with another they
are plainly present. Therefore, the feature themselves depend on how the form is
organized by the viewer.
In reality, neither type of processing “goes first”, instead both work together, with the
result that the perception that is achieved makes sense at both the large-scale & fine-
Recognition: Some Early Considerations
- You can recognice objects, various actions, & different sorts of situations. You can also
recognize variations of these things, as well as partial versions of these things (e.g. you
can identify a cat from it’s head and a paw, or identify a chair even if someone is sitting
- The same is true for print as well: you can recognize words printed in large, small,
italics, upper case or lower case.
- Recognition is also influenced bu the context of the objects encountered.
- Bottom-up influences: (stimulus driven) influences that come from the features in view
- Top-down influences: (knowledge/expectation drive) influences that come from you, in
which you supplement the input with your broader knowledge.
- Recognition might begin with the identification of visual features in the input pattern
(lines, curves, etc.) allowing you to assemble them into larger units.
Focusing on features could explain why we can recognize different letters in different
fonts, sizes, etc. (by noticing commonalities)
- People are fast at search for a target defined by a simple feature (vertical line in a field
of horizontals). People are much slower in searching for a target defined as a
combination of features.
- Integrative agnosia: (damaged parietal cortex) can detect particular features in a display,
but are impaired when it comes to judging how the features are bound together to form
Similar pattern noticed using TMS on parietal lobe
Factors Influencing Recognition
- Tachistoscopic presentations: show stimuli for a precisely controlled amount of time - Each stimulus followed by a post-stimulus mask (random jumble of letters) that
interrupts any continued processing of the stimulus.
- People recognize briefly visible stimuli depending on how familiar it is. Familiarity of
words can be measured by how often it appears in print.
- If people view a word & then view it again shortly after, they will recognize it more
readily the second time. The first exposure primes the participant – this is a case of
The Word-Superiority Effect
Words are easier to perceive than isolated letters
- “two-alternative, forced-choice” procedure: flash 1 letter (K) then ask whether E or K
was shown. Flash 1 word (DARK) then ask whether E or K was shown. Accuracy rates
were higher in the word condition.
Degrees of Well-Formedness
- Letters in words that are not English but look like English & are easy to pronounce (e.g.
FIKE) are easier to identify than letters alone.
- When asked to describe what they’ve seen, participants find it very difficult to
recognize random letters, but are more likely to identify words like FIKE or LARE.
Easily pronouncable strings do provide a context benefit & are easier to recognize.
The more statistically English-like the string, the easier to recognize & the greater the
- There is a strong tendency to misread less-common letter sequences as if they were
more-common patterns: TPUM likely misread as TRUM or DRUM or even something as
lengthy as TRUMPET. But the reverse if very rare: DRUM read as TRUM.
- Misspelled words, partial words, or nonwords are read in a way that brings them in line
with normal spelling; perceive input as more regular than it actually is.
Feature Nets & Word Recognition
The Design of a Feature Net - The idea, then, is that there is a network of detectors, organized in layers, with each
subsequent layer concerned with more complex, larger-scale objects.
Bottom-up – from lower levels toward upper levels.
- Each detector has a particular activation level, which reflects roughly how energized the
detector is. A strong input will increase the activation level by a lot, & so will a series of
weaker inputs. When the activation level reaches the detector’s response threshold, it
with fire – sending a signal to its connecting detectors.
- Detectors will require different strengths of input, in part due to how activated a
detector is to begin with. Detectors that have fired recently (“warm up”) or that have fired
frequently in the past (“exercise”) have a higher activation level Therefore activation
level is dependent on recency & frequency.
- Because frequent words appear more often, the detectors needed to recognize those
words have been used frequently & therefore have high levels of activation. Thus, even a
weak signal will bring these detectors to their response threshold & cause them to fire.
- Also explains repetition priming, because the relevant detectors will be “warmed up”
the second time.
The Feature Net & Well-Formedness
- Bigram detectors: detectors of letter pairs. - e.g. HICE – HI is familiar because it is in a lot of words