Cognition - Exploring the science of the mind
1. The foundations of cognitive psychology
In the late 19th century, Wilhelm Wundt and his student Titchener launched the new enterprise of
research psychology, defining their field for the first time as an endeavor separate from philosophy
or biology. In their view, psychology needed to be concerned largely with the study of conscious
mental events (feelings, thoughts, perception, and recollections). They concluded therefore that
the only way to study thoughts if for each of us to introspect, or look within, to observe and record
the content of our own mental lives and the sequence of our own experiences. But as a scientific
tool, introspection is worthless. Instead, psychology needed objective data, and that meant
research needed to focus on data that were out in the open, for all to observe. First, an organism’s
behaviors are obviously observable in the right way. My learning history can also be objectively
recorded and scientifically studied. We need to rule out any discussion of mentalistic notions. It
was this perspective that led researchers to the behaviorist movement. In brief: we need to study
the mental world, but we can’t. To use Kan’t transcendental method, you begin with the observable
facts and then work backward from those observations.
We know that we need to study mental processes; that’s what we learned from the limitations of
behaviorism. But we also know that mental processes cannot be observed directly; we learned
that from the downfall of introspection. Our path forward, therefore, is to study mental processes
indirectly, relying on the fact that these processes, themselves invisible, have visible
consequences: measurable delays in producing a response, performances that can be assesses
for accuracy, errors than can be scrutinized and categorized. By examining these (and other)
effects produced by mental processes, we can develop - and then test - hypotheses about what
the mental processes must have been = Kant.
3. Recognizing objects
How do you manage to perceive, and then recognize, the objects you see every day? This is the
problem, first, of form perception, the process through which you manage to see the basic shape
and size of an object. Next is the problem of object recognition, the process through which you
identify what the object is. Object recognition is crucial for learning. A group called the Gestalt
psychologists told that our perception of the visual world is organized in ways that the stimulus
input is not. They argued therefore that the organization must be contributed by the perceiver; this
is why, they claimed, the perceptual whole is often different from the sum of its parts. Your
perception is guided by principles of proximity and similarity. Each of us imposes our own
interpretation on the perceptual input, but we all tend to impose the same interpretation, because
we’re all governed by the same rules. In thinking about our discussion so far, it seems plausible
that perception proceed two broad steps: First, we collect information about the stimulus. Then,
once we’ve gathered the raw data, we interpret this information, and that’s when we go beyond the
information given. However, this view is wrong, and, in fact, our interpretation of the input
sometimes seems to happen before we start cataloguing the input’s basic features, not after.
Neither type of processing goes first. Neither has priority. Instead, both work together, with the
result that the perception that is achieved makes sense at both the large-scale and fine-grained
levels.
,Some influences come directly from the stimulus itself, the features that are in view. These
influences are sometimes called stimulus driven but more commonly are termed bottom up
influences. Other influences come from you, rather than the stimulus itself. These influences -
relying on your knowledge - are sometimes called knowledge driven or expectation driven, but are
more commonly called top down influences. Several lines of evidence indicate that object
recognition does begin wight he detection of simple features. Then separate mechanisms are
needed to put the features together, assembling them into complete objects.
To study word recognition, investigators often use tachistoscopic presentations. In these studies,
wordt that appear frequently in the language are easier to identify, and so are words that have
been recently viewed - an effect known as repetition priming. The data also show a pattern known
as the word-superiority effect; this refers to the fact that words are more readily perceived than
isolated etters. In addition, well-formed non words are more readily perceived than letter strings
that do not conform to the rules of normal spelling. Another reliable pattern is that recognition
errors, when they occur, are quid systematic, with the input typically perceived as being more
regular than it actually is. These findings together indicate that recognition is influenced by the
regularities that exist in our environment. These results can be understood in terms of a network of
detectors. Each detector collects input and fires when the input reaches a threshold level. A
network of these detectors can accomplish a great deal; for example, it can interpret ambitious
inputs, recover from its own errors, and make inferences about barely viewed stimuli. The feature
net seems to know the rules of spelling abed expects the input to conform to these rules. However,
this knowledge is distributed across the entire network and emerges only through the network’s
parallel processing. This setup leads to enormous efficiency in our commerce with the world
because it allows us to recognize patterns and objects with relatively little input and under highly
diverse circumstances. But these gains come at the cost of occasional error. This trade-off may be
necessary, though, if we are to cope with the information complexity of our world.
Stimulus input —> feature detectors —> letter detectors —> diagram detectors —> word detectors
The network does make mistakes, misreading some inputs and misinterpreting some patterns.
These errors are produced by exactly the same mechanisms that are responsible for the network’s
main advantages - its ability to deal with ambiguous inputs, for example, or to recover from
confusion. Over he years, researchers have offered variations on this basic conceptualization. All
three preserve the basic idea of a network of interconnected detectors, but all three extend this
idea in important ways:
- In the network proposal we’ve considered so far, activation of one detector serves to activate
other detectors. Other models include this notion of spreading activation but also include the
possibility of detectors inhibiting each other, so that activation of a detector can in fact serve to
decrease the activation in other detectors. The McClelland and Rumelhart model is better able to
identify well-formed strings than irregular strings; this net is also more efficient in identifying
characters in context as opposed to characters in isolation. However, several attributes of this net
make it possible to accomplish all this without bigram detectors. Some of the connections are
excitatory - so that activation of one detector causes activation in its neighbors (excitatory
connections). Other connections are inhibitory, and deactivates another detector (inhibitory
connections). Higher-level detectors can influence the lower-lever detectors, and detectors at
ably level can also influence other detectors at the same level.
, - The recognition by components model (RBC) includes several important innovations, one of
which is the inclusion of an intermediate level of detectors, sensitive to geons (geometric ions).
The idea is that geons might serve as the building blocks of all the objects we recognize; geons
are, in essence, the alphabet from which all objects are constructed. Geons are simple shapes,
such as cylinders, cones, and blocks. The RBC model uses a hierarchy of detectors. The
lowest-level detectors are feature detectors, which respond to edges, curves, and so on. These
detectors in turn activate the geon detectors. Higher levels of detectors are then sensitive to
combinations of geons (geon assemblies), which explicitly represent the relations between
geons. These assemblies, finally, activate the object model, a representation of the complete,
recognized object.
- A number of researchers propose that people have stored in memory a number of different
views of each object they can recognize. The key then is that recognition sometimes requires
mental rotations, and, as a result, will be slower from some viewpoints than from others.
There is a special neural structure involved almost exclusively in the recognition and discrimination
of faces, it’s this structure that is damaged in people suffering from prosopagnosia. Face
recognition is also specialized in another way - in its strong dependence on orientation.
Prosopagnosia is not strictly a disorder of face recognition! Face recognition doesn’t depend on an
inventory of a face’s parts: instead, this recognition seems to depend on holistic perception. The
features matter by virtue of the relationships and configurations they create. It’s these
relationships, ant not the features on their own, that guide face recognition.
4. Paying attention
It seems attention may be needed for conscious perception. The active nature of perception is
evident in studies of change blindness - observers’ inability to detect changes in scenes they’re
looking at directly. According to the early selection hypothesis the attended input is identified and
privileged from the start, so that unattended input receives little analysis. According to the late
selection hypothesis all inputs receive relatively complete analysis, and the selection is done after
all of this analysis is finished Perhaps the selection is done just before the stimuli reach
consciousness, and so we become aware only of the attended input (relatively difficult input). Or
perhaps the selection is done later still - so that all inputs make it into consciousness, but then the
selection is done, so that only the attended input is remembered (relatively simple input). It turns
out that each hypothesis captures part of the truth. The proposal is that people can literally
prepare themselves for perceiving by priming the suitable detectors. Thanks to prior exposure, the
activation level of detectors are already high; you don’t need to prime them further. You don’t do
this priming for inputs you have no interest in, and you ca’t do this priming for inputs you can’t
anticipate. We need to distinguish two types of primes. One type is stimulus based - produces
merely by presentation of the priming stimulus, with no role for expectation. The other type is
expectation-based and is created only when te participant believes the prime allows a prediction of
what’s to come. These types of primes can be distinguished in various ways. First, expectation-
based priming takes longer to kick in that stimulus-based priming. Stimulus=based priming
appears to be free, and so we can prime one detector without taking anything away from the other
