Perception summary
FOUNDATIONS & PSYCHOPHYSICS
How should we study perception? → Perception is fundamental to our life.
- Protagoras (490-420 BC): “Man is the measure of all things: of things we are, that they are, and of
things which are not, that they are not.”
- physical reality vs perceptual reality → differences in experiences and what you perceive
- there are many things we cannot actually perceive, because our senses don’t or we are not
sensitive to it
Is our perceptual system the perfect measuring device that we think it is? Can we trust our
perceptual reality?
- Kant (1742-1804 AD): “[…] the senses do not err – not because they always judge rightly, but
because they do not judge at all.”
- Fraser spiral illusion: it appears to be a spiral, but they are perfect circles → it looks like that there
is just one line and that it is a
kind of a spiral, but when you
follow the line with a pen or
something, you see that there is
no spiral
The perceptual process
- the dissociation between
sensation and perception isn’t
very strict: there could be
sensation-like properties also
relevant for more cognitive
processing
- three main types of questions:
1. How does the proximal stimulus carry information about the thing that is perceived (i.e., the distal
stimulus)?
2. How is the proximal stimulus transformed into neural signals?
3. What is the relationship between perceptual experience and the distal stimulus, the thing that is
perceived?
- bottom-up processes are driven by the world out there
- top-down processes are things like expectations that influence our perception
(4. How does the signal processing in the brain go?)
(5. Conscious awareness)
The senses
- BUT also proprioception (sense of self-movement and body position), pain perception,
thermoreception, balance, body movement
, Primary Physical Transduced Sensed physically Sensed perceptually
sense stimuli receptors
Vision Light Photoreceptors in Intensity, Brightness, color,
(sight) eye wavelength, shape, texture,
spatiotemporal location, motion
distribution
Audition Sound Hair cells in inner Amplitude, Loudness, pitch,
(hearing) ear (in cochlea) frequency, waveform, timbre, location of
interaural differences sound source
Tactile Mechanical Mechanoreceptors Skin indentation, skin Spatial pattern,
perception forces in skin stretch, skin motion, texture, shape, hand
(touch) vibration of skin conformation, slip,
grip control, fine
texture, transmitted
vibration
Olfaction Molecules Olfactory receptor Molecular structure Odor
(smell) neurons in nose and concentration
Gustation Molecules Taste receptor Molecular structure Sweetness, saltiness,
(taste) cells in mouth and concentration bitterness, sourness,
and others
Evolution has played a large role in the development of our perceptual system → huge variety in
eyes → (slightly) different functions, different sensitivities (predator/prey)
- we have the features that we have due to the evolutionary process
Measuring perception
- for measuring behavior: psychophysics
- first person data: something that is in your personal head (what is the colour red to you
perceptually?)
- first vs. third person data: to assess objectively the relationship between stimuli and experience
perception can differ between people, even when the stimulus is the same
- you need some kind of measure
- topics of psychophysics:
- thresholds: minimum intensity and minimum difference (what is the minimum intensity
that we need to perceive a particular stimulus? What is the minimum differences between
the two lights that we can perceive?)
- scaling: how does our perceptual experience change with differences in the stimulus in the
environment?
Finding the absolute threshold (detection) → methods:
1. Method of constant stimuli
a. You show stimuli of different intensities, and do several trials
b. You detect at what intensity the stimulus was
perceived, you do several trial
c. You take the average of all trials
- drawback to the system (quite laborious): you
have to do a lot of presentations of the different
intensities
2. Method of limits / method of adjustment
a. Look for threshold in a slightly different way →
,use fewer stimuli
b. You have a more methodological approach → certain sequence (no random presentation) → at
what point the participant’s response changes
c. Once you reached the threshold, you stop presenting the stimuli after that threshold. The next
trial, you start at the other side (light to dark vs. dark to light) until the response changes again (and
then switch again)
d. Function of perception is S-shaped. The threshold is when 50% of the people perceive the stimulus
The psychometric function: S-shaped function, responses are bit more graded (don’t reach the 100%
immediately after the threshold)
- gradient function
- absolute threshold: half/half chance that you see (perceive) it → 50% (above threshold: more likely
to see it, beneath threshold: less likely to see it)
3. Staircase method
a. Start with first trial and stop with this trial when the response is changed
b. You start the next trial at 1 below where change occurred
c. Drawback of system: this observation is not fully complete because you
don’t look at all intensities (what perceives/does the person at the high
intensities?) → psychometric function isn’t complete
- So what do we find then?
we can detect light of a single candle 48 km away, at night
we can hear from 6 m away a ticking watch, in a silent room
we can taste a tablespoon of salt in 7.5 L of water
we can feel a fly wing dropped from 7.6 cm height, on your
cheek
Signal detection theory (SDT)
- used because people are very unreliable (problem in research), they
do crazy things
- how do yes/no responses come to be? → decision strategies (bias (someone says that he can see
everything (not reliable)) next to just random noise (sensitivity)
- neural noise = neural response to a stimulus, BUT there is almost always some neural firing, even
when there is no clear stimulus
→ also neural firing in absence
of stimulus (no 0)
- normal distribution: there is a
range of firings that is
associated with the
onset/presence of that
stimulus
- target-present vs. target
absent signals (e.g. spikes)
- baseline firing = the spikes
when no tone is presented
(shows a signal absent
distribution)
- criterion = value set at the
point at which only the
, stimulus is responsible for the neural firing (spikes) → exclude the baseline firing (target absent
signals) and look at the number of spikes that occur as a response to the (present) stimulus
But… what if the criterion value varies? → your chosen criterion value can vary, e.g. when you’re
looking for weapons in luggage → you can better have a false
alarm sometimes, to make sure you never miss one
- catch trials allow separation of actual sensitivity (d’) and decision
making bias (criterion: β)
The criterion visualized on a Receiver Operating Characteristic
(ROC) plot
- changing the decision criterion: decreasing the criterion leads to a higher hit rate, but also a higher
false alarm rate (trade off!)
- arbitrary decision
Sensitivity visualized on an ROC
plot:
- d’ = sensitivity
- increasing d’ leads to a higher hit
rate and a lower false alarm rate
- d’ = 0 → 50% hit, 50% false alarm
(just guessing)
- lower d’ → trade-off is far more
severe
Back to finding thresholds: the
difference threshold
(discrimination), where you can
discriminate one stimulus to
another
- also known as the just noticeable
difference (JND): area between
25% of yes-responses and 75%,
FOUNDATIONS & PSYCHOPHYSICS
How should we study perception? → Perception is fundamental to our life.
- Protagoras (490-420 BC): “Man is the measure of all things: of things we are, that they are, and of
things which are not, that they are not.”
- physical reality vs perceptual reality → differences in experiences and what you perceive
- there are many things we cannot actually perceive, because our senses don’t or we are not
sensitive to it
Is our perceptual system the perfect measuring device that we think it is? Can we trust our
perceptual reality?
- Kant (1742-1804 AD): “[…] the senses do not err – not because they always judge rightly, but
because they do not judge at all.”
- Fraser spiral illusion: it appears to be a spiral, but they are perfect circles → it looks like that there
is just one line and that it is a
kind of a spiral, but when you
follow the line with a pen or
something, you see that there is
no spiral
The perceptual process
- the dissociation between
sensation and perception isn’t
very strict: there could be
sensation-like properties also
relevant for more cognitive
processing
- three main types of questions:
1. How does the proximal stimulus carry information about the thing that is perceived (i.e., the distal
stimulus)?
2. How is the proximal stimulus transformed into neural signals?
3. What is the relationship between perceptual experience and the distal stimulus, the thing that is
perceived?
- bottom-up processes are driven by the world out there
- top-down processes are things like expectations that influence our perception
(4. How does the signal processing in the brain go?)
(5. Conscious awareness)
The senses
- BUT also proprioception (sense of self-movement and body position), pain perception,
thermoreception, balance, body movement
, Primary Physical Transduced Sensed physically Sensed perceptually
sense stimuli receptors
Vision Light Photoreceptors in Intensity, Brightness, color,
(sight) eye wavelength, shape, texture,
spatiotemporal location, motion
distribution
Audition Sound Hair cells in inner Amplitude, Loudness, pitch,
(hearing) ear (in cochlea) frequency, waveform, timbre, location of
interaural differences sound source
Tactile Mechanical Mechanoreceptors Skin indentation, skin Spatial pattern,
perception forces in skin stretch, skin motion, texture, shape, hand
(touch) vibration of skin conformation, slip,
grip control, fine
texture, transmitted
vibration
Olfaction Molecules Olfactory receptor Molecular structure Odor
(smell) neurons in nose and concentration
Gustation Molecules Taste receptor Molecular structure Sweetness, saltiness,
(taste) cells in mouth and concentration bitterness, sourness,
and others
Evolution has played a large role in the development of our perceptual system → huge variety in
eyes → (slightly) different functions, different sensitivities (predator/prey)
- we have the features that we have due to the evolutionary process
Measuring perception
- for measuring behavior: psychophysics
- first person data: something that is in your personal head (what is the colour red to you
perceptually?)
- first vs. third person data: to assess objectively the relationship between stimuli and experience
perception can differ between people, even when the stimulus is the same
- you need some kind of measure
- topics of psychophysics:
- thresholds: minimum intensity and minimum difference (what is the minimum intensity
that we need to perceive a particular stimulus? What is the minimum differences between
the two lights that we can perceive?)
- scaling: how does our perceptual experience change with differences in the stimulus in the
environment?
Finding the absolute threshold (detection) → methods:
1. Method of constant stimuli
a. You show stimuli of different intensities, and do several trials
b. You detect at what intensity the stimulus was
perceived, you do several trial
c. You take the average of all trials
- drawback to the system (quite laborious): you
have to do a lot of presentations of the different
intensities
2. Method of limits / method of adjustment
a. Look for threshold in a slightly different way →
,use fewer stimuli
b. You have a more methodological approach → certain sequence (no random presentation) → at
what point the participant’s response changes
c. Once you reached the threshold, you stop presenting the stimuli after that threshold. The next
trial, you start at the other side (light to dark vs. dark to light) until the response changes again (and
then switch again)
d. Function of perception is S-shaped. The threshold is when 50% of the people perceive the stimulus
The psychometric function: S-shaped function, responses are bit more graded (don’t reach the 100%
immediately after the threshold)
- gradient function
- absolute threshold: half/half chance that you see (perceive) it → 50% (above threshold: more likely
to see it, beneath threshold: less likely to see it)
3. Staircase method
a. Start with first trial and stop with this trial when the response is changed
b. You start the next trial at 1 below where change occurred
c. Drawback of system: this observation is not fully complete because you
don’t look at all intensities (what perceives/does the person at the high
intensities?) → psychometric function isn’t complete
- So what do we find then?
we can detect light of a single candle 48 km away, at night
we can hear from 6 m away a ticking watch, in a silent room
we can taste a tablespoon of salt in 7.5 L of water
we can feel a fly wing dropped from 7.6 cm height, on your
cheek
Signal detection theory (SDT)
- used because people are very unreliable (problem in research), they
do crazy things
- how do yes/no responses come to be? → decision strategies (bias (someone says that he can see
everything (not reliable)) next to just random noise (sensitivity)
- neural noise = neural response to a stimulus, BUT there is almost always some neural firing, even
when there is no clear stimulus
→ also neural firing in absence
of stimulus (no 0)
- normal distribution: there is a
range of firings that is
associated with the
onset/presence of that
stimulus
- target-present vs. target
absent signals (e.g. spikes)
- baseline firing = the spikes
when no tone is presented
(shows a signal absent
distribution)
- criterion = value set at the
point at which only the
, stimulus is responsible for the neural firing (spikes) → exclude the baseline firing (target absent
signals) and look at the number of spikes that occur as a response to the (present) stimulus
But… what if the criterion value varies? → your chosen criterion value can vary, e.g. when you’re
looking for weapons in luggage → you can better have a false
alarm sometimes, to make sure you never miss one
- catch trials allow separation of actual sensitivity (d’) and decision
making bias (criterion: β)
The criterion visualized on a Receiver Operating Characteristic
(ROC) plot
- changing the decision criterion: decreasing the criterion leads to a higher hit rate, but also a higher
false alarm rate (trade off!)
- arbitrary decision
Sensitivity visualized on an ROC
plot:
- d’ = sensitivity
- increasing d’ leads to a higher hit
rate and a lower false alarm rate
- d’ = 0 → 50% hit, 50% false alarm
(just guessing)
- lower d’ → trade-off is far more
severe
Back to finding thresholds: the
difference threshold
(discrimination), where you can
discriminate one stimulus to
another
- also known as the just noticeable
difference (JND): area between
25% of yes-responses and 75%,
