Subliminal Perception

Zoltn Dienes Conscious and unconscious mental processes

Subliminal perception = perceiving without being aware of perceiving

Conscious perception can be blocked with masking

Masked stimulus Forward mask ISI Backward mask

ISI = inter-stimulus interval

time
SOA

SOA = stimulus onset asynchrony

Monoptic: mask and word presented to one eye the same eye

Dichoptic: mask and word presented to different eyes

Noise mask must be presented to same eye as tobe-masked stimulus MXgFhhY

Noise mask:
Each pixel can be randomly on or off

Pattern mask:

Pattern mask can be presented dichoptically

Random collection of the parts of the stimulus to be masked e.g. letters or letter parts for word stimuli

=> The effect of a pattern mask on visual processing occurs more centrally than the effect of noise masking

Marcel (1983) Experiment Four

N=12 subjects , three masking conditions

RT (msec)

Unassocd

Associated

Facilitation

Unmasked control
Pattern masked (dichoptic) Noisemasked (monoptic)

590

528

+62

597

541

+56

586

582

+4

Cheesman & Merikle (1984, 1986)

In the Marcel experiments, if subjects not confident they definitely saw something, they may say no on every trial.
=> discrimination performance = 50%

But if subjects are forced to respond yes 50% of time, maybe they would be above chance?
Maybe Marcel had not found the objective threshold for discrimination?

Cheesman & Merikle presented subjects with one of four colour words (red, green, etc). On each trial, subjects had to say which word had been presented. (Now even if subjects believed they saw nothing, have to guess something.)

After each block of 40 trials, the SOA was reduced until subjects were performing at chance. After each block of trials, subjects estimated how accurate they had been. - If they felt they had no information whatsoever, they were just guessing purely randomly, subjects gave an estimate of 25% correct (=chance expectation). - If they were certain, they would give an estimate of 100%.

- Subjects could give any value between 25% and 100%.

Subjective threshold = the SOA at which subjects believe they are performing at chance Objective threshold = the SOA at which subjects really are performing at chance.

Results: 1) The subjective threshold is reached at a higher SOA than the objective threshold. => the two thresholds are different 2) Stroop priming is obtained for colour words presented below the subjective threshold, but NOT below the objective threshold.

Conclusion: BOTH results indicate unconscious processing below a subjective threshold.

But no evidence for unconscious processing below an objective threshold.

Research since then has confirmed it is easy to get unconscious perception below a subjective threshold, but difficult below an objective threshold (for semantic processing)

Greenwald (1992) found that 93% of the cognitive psychologists he surveyed regarded subliminal perception as having been demonstrated below a subjective threshold.

Process dissociation applied to unconscious perception

Jacoby, Toth, Lindsay, & Debner (1992) Procedure:

time 1000 ms 500 ms 50 or 500 ms 'glove'

event ' +'

'patch' or 'staff'

500 ms
500 ms response

'flare'

pat--

Exclusion test: Complete the stem with a word that comes to mind but not any you just saw displayed. Inclusion test: Complete the stem with one of the words flashed or, if unable to do so, with the first word that comes to mind. Another way of testing for the subjective threshold you exclude a word if you think you saw it.

Results: 500ms 50ms Baseline

Exclusion
Inclusion

.10
.96

.50
.63

.36
.38

For 50 ms words:
Any difference between inclusion and exclusion must be based on conscious perception:

Prob. of conscious perception = .13

Results: 500ms 50ms Baseline

Exclusion
Inclusion

.10
.96

.50
.63

.36
.38

Conscious perception would make exclusion performance BELOW baseline

At 500ms, exclusion is indeed below baseline

At 50ms, exclusion is ABOVE baseline, so there must have been unconscious perception

Inattentional blindness: Information can be below a subjective threshold when attention is focussed on another object Mack and Rock 1998 Which arm of the cross is longer horizontal or vertical? 500 ms 200 ms 1500 ms 60% subjects reported not seeing the word on the third trial Of these blind subjects, 36% completed a word stem with the presented word (4% for control). Recognition also above chance.

Is the subjective threshold just a curiosity or theoretically interesting? Need to establish qualitative differences between knowledge above and below the subjective threshold. Merikle & Joordens (1997) Only two words used: Red/green. Red or green

(variable SOA)
backward mask &&&&& (in red or green ink)

Task: Name colour of ink. 75% of trials, prime and target incongruent (e.g. red followed by green ink) ; 25% of trials congruent

400

390

380

TRIAL
Mean RT
cong 370 sub supra incong

THRESH

A conscious belief (e.g. I am seeing the word RED) can be combined with any other belief or desire to which it may be relevant in order to produce further beliefs or actions (in this case, to be prepared for green ink). Unconscious knowledge cannot combine with just any other possibly relevant belief or desire to plan action. => Only conscious knowledge is inferentially promiscuous.

Semantic processing under the objective threshold? Naccache et al 2005 Patients undergoing brain surgery categorised masked and unmasked words as threatening or neutral

The three patients at chance in categorising the masked words

Electric field potentials recorded in the amygdala to threatening (red) and neutral (green) words. Blue line = significant difference

Both conscious and subliminal words lead to different amygdala activations in all three patients

Neural correlates of consciousness What is the difference between conscious and unconscious perception at the level of wetware? What process in the brain is correlated with conscious awareness? What brain structure or process corresponds to having higher order thoughts?

Strategies using brain imaging, e.g. fMRI: 1. Compare conscious perception with no conscious perception when external stimulus is the same (binocular rivalry) 2. Compare conscious and unconscious perception of the same stimulus (conscious vs subliminal perception)

strategy 1. Binocular rivalry One eye presented with e.g. face, the other with e.g. a house Both stimuli constantly presented but one only sees one thing at a time - conscious percept alternates every few seconds What brain activity correlates with the alternating conscious experience?

FFA = fusiform face Area; PPA = parahippocampal place area

Leopold & Logothetis, 1996

Activity in V1 not related to conscious perception.

Lumer & Rees (1999): Conscious perception not just related to activity in specific visual areas like fusiform also to prefrontal cortex which is involved in working memory and planning.

Strategy 2: compare brain activity between conscious and unconscious perception of the same stimulus Dehaene et al 2001 fMRI words either at objective threshold or clearly visible:

Both conscious and unconscious words activated left fusiform area (to some degree); conscious words led to more extensive brain activity, including the prefrontal cortex.

Allows two interpretations:

1. Conscious awareness depends simply on information being widely broadcast in the brain (Baars, Dennett, Dehaene)

2. Conscious awareness depends on information reaching a certain area, e.g. prefrontal cortex (the HOT box?)

Lau and Passingham 2006: Need to equate objective performance between conscious and unconscious and only subjective experience differ - otherwise performance confounded with consciousness Two conditions: Visual discrimination task with same level of objective performance but different probability of thinking one saw the stimulus

Activation difference between conscious and unconscious now very specific: left Mid dorsolateral prefrontal cortex HOT box responsible for creating accurate higher order thoughts?