Finally, we ran an extra experiment where we entirely removed the detection component from the setup and show that the base rate effect disappears, whereas the attention effect remained. 9/11

So we fit a Hurdle-Gaussian model that allows us to separate the decision and the perceived contrast components that inform reproduction responses. Attention moved the contrast. Base rate and payoff mostly moved the hurdle (decision). These results were replicated in a new sample. 8/11

When looking deeper, we figured that reproduction responses are the combination of how often people set the patch to zero (saw nothing), and how bright they set it when they do see something. Very different mixes of the two give you the exact same average. 7/11

Interestingly, reproduced contrast went up under all three as well. Taken at face value that would mean statistical priors and reward structures literally change what you consciously see. We were suspicious. 5/11

We pushed people toward "present" responses in three ways: an attentional cue, an asymmetrical base rate (90% of trials had a target), and a payoff where misses cost 5x more than false alarms. All three biased detection responses. 4/11

Across three experiments, we asked 505 participants to detect very dim Gabor patches, sometimes present, sometimes not, or to reproduce what they saw by adjusting the patch's contrast. Crucially, participants did not know which task they were doing until after the stimulus was gone. 3/11

When directly comparing prime and mask discrimination, we found an indirect task advantage in the same three participants for whom we showed unc. priming using the double t-test approach.

Even after training, some SOAs still showed chance prime discrimination (although not in all SOAs, as in Vorberg et al.). Using the double t-test approach, as in Vorberg et al. (2003), we established unconscious priming in at least one SOA for 3 participants.

Unlike in Vorberg et al., some participants could already discriminate the prime at longer SOAs even before training. In the priming task, however, we replicated Vorberg’s main pattern of results.

In Vorberg et al. (2003), participants were slower to respond to an arrow-shaped mask when it followed an arrow-shaped prime pointing in the opposite direction, despite participants not being aware of the prime. Or they were faster if both prime and mask were pointing in the same direction.

I will be presenting some of my postdoc work during the very last session of the conference. would love to hear your feedback 📊📢 @assc28.bsky.social #ASSC28

(couldn't wait till Monday because I'm impatient) this lovely preprint has the last part of my phd work and now is out and about 🎉🎊. will write a thread on Monday. stay tuned 📻. osf.io/preprints/ps...

Further, aggregated measures of metacognition (M-ratio) were also biased by non-perceptual manipulations (as identified by the reproduction task).

The same applied when confidence was reported after first-order decisions.

While all manipulations affected first-order decisions and concurrent confidence reports, only the Muller-Lyer illusion influenced the reproduced length of the lines.

Participants discriminate long and short lines, indicate their conf. on the line being short or long (concurrently or after their first order decision), or to reproduce the length of the line. We use the Muller-Lyer illusion, a stim. prevalence manipulation and a payoff scheme to bias participants.

To celebrate the defence of my PhD, and thanks to @ibbamsterdam.bsky.social, I will host a symposium titled "Vision as prediction: learning, action, and biases" on June 6. Program, registration and more information here: bit.ly/43reOgR

are we about to get an eXistenZ VR console?

thanks to the iBBA, to all the people that assited to the workshop and specially to @robince.bsky.social and @canalesjohnson.bsky.social for their enriching and thought provoking lectures/tutorials

second day of the iBBA Info. Theory Workshop with @canalesjohnson.bsky.social

cat thread

thanks to the iBBA VU our workshop on Info. Theory is in full swing with Dr. Ince (@robince.bsky.social) giving an intro. lecture on info. theory measures. Dr. Canales-Johnson (@canalesjohnson.bsky.social) will follow with a lecture on how to use these measures to study perception and cognition

i need some empathetic comments 😪

very excited to be at the ASSC and even more excited to be in tokyo 🗼 #ASSC27 I will be presenting some of my postdoc work tomorrow at 14:45 in the Unconscious Processing session. drop by so I can tell you how we replicate some crucial unconscious processing findings.