Factors Influencing Recognition
o In many studies participants have been shown stimuli for a brief duration (20-30ms). In
older research this was done by means of tachistoscope, a device specifically designed
to present stimuli for precisely controlled amounts of time.
o Each stimuli is followed by a post-stimulus mask- randomly jumbled letters, which
serves to interrupt any continued processing. This way researchers can be certain that
the stimulus presented for only that specific amount of time.
o Can people recognize this stimulus? It depends on many factors including familiarity.
o Jacoby and Dallas showed participants words that were either very frequent or
infrequent, participants viewed these words for 35ms, followed by a mask; under these
circumstances, and they recognized twice as many frequent words as infrequent words.
o Recency of view: if participants view a word and then, a little later, view it again, they will
recognize the word much more readily the second time. The first exposure primes the
participant for the second exposure; more specifically, this is a case of repetition
The Word Superiority Effect
o Words themselves are easier to perceive, as compared to isolated letters.
o This effect is usually demonstrated with a “two-alternative, forced choice” procedure.
o In some trials a single letter will be presented while in others a full word will be presented
and then the participant will be asked to choose which letter was presented out of a
choice of two letters (50-50 chance).
o Accuracy rates are higher in the word condition.
Degrees of Well-Formedness
o The word superiority effect will only occur if the string of letters is a English word or a
word that resembles that of one (easy to pronounce).
o Pronounceability: easy pronounceable strings provide a context benefit (i.e. FIKE).
o “Englishness” is a good predictor of word recognition.
o Spelling patterns. Chapter 3
o With brief exposures, word recognition is good but not perfect, and the errors that occur
are quite systematic: the strong tendency to misread less common letter sequences as if
they were more common patterns; irregular patterns are misread as regular patterns
(“TPUM” is misread a “TRUM” or “DRUM”).
o Large errors may also occur. (“TPUM” being misread as “TRUMPET”).
o Misspelled words, partial words, or none words are read in a way that brings them into
line with normal spelling.
Feature Nets and Word Recognition
The Design of a Feature Net
o There is a network of detectors, organized in layers, with each subsequent layer
concerned with more complex, larger-scale objects.
o The “bottom” layer is concerned with features, and that is why networks of this sort are
often referred to as feature nets, and, the flow of information would be bottom-up.
o Each detector in the network has a particular activation level, which reflects the status of
the detector at just that moment.
o When a detector receives some input, its activation level increases.
o The activation level will eventually reach the detector’s response threshold, and at that
point the detector will fire (sending a signal to the other detectors it is connected with).
o If the feature net is to be a serious candidate for how humans recognize patterns, then it
has to use the same sort of building blocks that the brain does. Detectors likely involve
complex assemblies of neural tissues (function in the way that’s biologically sensible).
o Some detectors require a strong input to make them fire, while others will fire even with
a weak input. This difference is created in part by how activated each detector is to begin
with. If the detector is moderately activated to begin with it will only need a little input to
raise the activation level to the response threshold, and therefore it will be easier to
make the detector fire.
o Activation level is dependent on the principles of recency and frequency.
o Repetition priming; only a weak signal will be needed to make the detector fire again. As
a result, the word will be more easily recognizable the second time around.
The Feature Net and Well-Formedness
o If none of the letter sequences create a word, the word detectors will play no role in the
recognition of the strings. Chapter 3
o Adding another level to the net accommodation detectors for letter combinations.
o Bigram Detectors: detectors of letter pairs. These detectors, like the rest, will be
triggered by lower-level detectors and send their output to higher-level detectors.
o When none of the letter combinations are familiar, the string will receive no benefits from
priming. A strong input will therefore be needed to bring the relevant detectors to
threshold, and so the string will be recognized only with difficulty.
Recovery from Confusion
o With a brief presentation (20ms), the visual system has only a limited opportunity to
analyse the input, so it’s possible that the visual system will miss some of the features
that are present.
o This partial information invites confusion: many letters share common forms (i.e. bottom
o Activation in the feature detector causes activation in all letter detectors.
o However, each letter detector is wired so that it also receives input from other feature
o We have partial information at the feature level, leading to the confusion at the letter
level. The information send upward, from the letter level to the bigram level, reflects this
o This confusion is then sorted out at the bigram level. The multiple bigram detectors in
the situation are all retrieving the same input-a strong signal from one of their letters and
a weak one from the other. They don’t all react the same way. If the bigram detector is
well primed (more frequent) the more likely it is to fire.
o Context influe