The authors have declared that no competing interests exist.
Conceived and designed the experiments: JS. Performed the experiments: JS. Analyzed the data: JS MM. Contributed reagents/materials/analysis tools: JS MM. Wrote the paper: JS MM.
The effects of novelty on low-level visual perception were investigated in two experiments using a two-alternative forced-choice tilt detection task. A target, consisting of a Gabor patch, was preceded by a cue that was either a novel or a familiar fractal image. Participants had to indicate whether the Gabor stimulus was vertically oriented or slightly tilted. In the first experiment tilt angle was manipulated; in the second contrast of the Gabor patch was varied. In the first, we found that sensitivity was enhanced after a novel compared to a familiar cue, and in the second we found sensitivity to be enhanced for novel cues in later experimental blocks when participants became more and more familiarized with the familiar cue. These effects were not caused by a shift in the response criterion. This shows for the first time that novel stimuli affect low-level characteristics of perception. We suggest that novelty can elicit a transient attentional response, thereby enhancing perception.
Detection of new information is crucial for adapting behavior to the present circumstances - therefore the human brain is tuned towards novelty
There is another reason to expect novelty to affect perceptual processing. Emotional stimuli are known to enhance early perceptual processing (for a review see
To investigate this hypothesis we investigated novelty’s effects on visual perception in a cueing paradigm. We did this by manipulating the novelty versus familiarity of cue stimuli that were all fractals. Every participant was familiarized with one of the fractals by viewing this stimulus for at least 20 seconds. In the subsequent blocks, 50% of cues consisted of this one familiar stimulus, familiarizing participants even more with it. The other category of stimuli, the novels, were all presented only once during the experiment to guarantee their novelty. The novel/familiar cues were presented centrally. The effects of the cues were measured in a tilt detection task. Performance on such a task is facilitated by contrast
In the first experiment we investigated the effects of novel cues on perception for tilts of different angles. In experiment 2, we elaborated on this design by measuring contrast sensitivity. In both experiments we predicted that novel cues would enhance sensitivity of perception of the target compared to familiar cues.
17 participants (12 female; 15 right-handed) participated in the experiment on a voluntary basis. All participants were naïve to the aim of the study and signed written informed consent before participation. The experiments were performed in accordance to the ethical standards laid down in the 1964 Declaration of Helsinki and approved by the faculty ethical committee, the Scientific and Ethical Review Board of the Faculty of Psychology and Education (VCWE). Participants were paid 6–7 Euros of compensation or were given course credits. The participants all had normal or corrected-to-normal vision.
The stimuli were presented on a 19 inch CRT monitor (1024x768 pixels) at a viewing distance of about 75 cm using E-Prime programming software (Psychology Software Tools Inc., Pittsburgh, PA, USA). The refresh rate of the screen was 120 Hz.
All stimuli were presented in the center of the screen, on top of a silver background (luminance CIE(.34,.39), 61.60 cd/m2). A trial started with a black fixation cross, followed by a cue display consisting of a centrally presented fractal image (10.2°×10.2°). Novel/familiar fractal stimuli were generated by iterative mathematical computations using the open-source program ChaosPro 4.0 (
For every participant one fractal was randomly chosen to become familiarized. This ‘familiar’ image was presented for 20 seconds prior to the experiment, in order for participants to become familiarized with it. The ‘novel’ images were all unique fractals unknown to the observer.
Before the experiments participants were told that accuracy on the task was most important, and that speeded responses were not required. In addition participants were instructed that a fixation cross would be presented between trials and stimuli, and that they should always fixate their eyes on this cross. All participants performed 48 training trials. Then 6 experimental blocks were completed of 128 trials, with a short break in the middle and at the end of each block. Half the trials had familiar cues, and half had novel cues. All 816 trials were performed in about 40 minutes.
Every trial started with the onset of a fixation cross presented for 1000 ms. After fixation a cue, either a novel or familiar fractal image, was briefly presented for 70 ms. Subsequently, the fixation cross was presented again for 30 ms. Then the target, consisting of a centrally presented Gabor patch (vertical/tilted), was presented for 70 ms. The timing of the stimuli was taken from a cueing paradigm used to investigate the effects of emotion on early vision
In Experiment 1 the target could be tilted either 0, 1, 2, 3, or 4 degrees. In Experiment 2 the target could be tilted with any of these angles, determined by the participant’s performance in the practice trials. In addition the contrast of Experiment 2 targets was varied in nine log increments.
We used a two-alternative forced-choice (2AFC) task in which participants pressed “x” to indicate that the target Gabor was oriented vertically, or “m” to indicate that it was tilted (either left or right). Visual feedback was given to correct (“Correct”) and incorrect (“Incorrect”) trials for 1000 ms after a response was given.
In addition, response bias was calculated for every condition and participant: β = -(Z(hit rate) – Z(false alarm rate))/2. Since there were no separate lure conditions for each angle, bias was only computed for the two cue type conditions and for different blocks. The same ANOVA as for
For every participant and condition hit rates and false alarm rates that exceeded.975 were clipped to.975 to prevent creating extreme outliers and allowing
Error bars reflect standard errors of the mean (between-subject).
In the ANOVA with cue type and experimental block as within-subject variables, there was again a main effect of cue type on
Familiar cues are often preceded by trials on which the same cue was shown. To investigate whether a possibly subsequent low-level adaptation to the familiar cue could have caused a reduction in the attentional resources allocated to the target, trials on which a familiar was preceded by either another familiar cue (mean
There was no main effect of cue type on response bias,
Error bars reflect standard errors of the mean (between-subject).
As predicted novel cues enhanced tilt detection compared to familiar cues. More specifically, orientation detection was improved when the target was preceded by a novel cue. This effect seemed to become stronger during the experiment, although statistical evidence for this remained at the level of a trend. The effect cannot be caused by forward masking of the target by the cue stimulus, since novel and familiar stimuli did not differ in stimulus properties. Familiar cues were often preceded by trials in which the same cue was used, while this was by definition never the case for the unique novel cues. Although it could theoretically lead to adaptation to familiar cues, repetition of the familiar cue had no effect on performance. This suggests that it is the familiarity or novelty of the cue, and not some short-range effect, that causes the differences found in the experiment.
No differences in response criterion were found, suggesting that novelty improved performance on the tilt detection task by enhancing sensitivity.
To replicate and extend the findings of Experiment 1, we performed a new experiment using a similar cueing task, now varying the contrast of the target Gabor patches.
17 participants (15 female; 15 right-handed) participated in the experiment on a voluntary basis. All participants were naïve to the aim of the study and signed written informed consent before participation. Participants were paid 9 Euros of compensation or were given course credits. The participants all had normal or corrected-to-normal vision.
The stimuli and apparatus were the same as in Experiment 1; however, now the participants were presented with Gabor patches from only one tilt angle (determined by practice block performance – see below). The task was identical to the one in Experiment 1, that is, participants had to indicate whether the briefly presented target Gabor patch was either tilted (button press “m”) or untilted (button press “x”). Accuracy, but not speed was emphasized. The new Gabor patches were created with a Matlab script, and had Michelson contrasts in nine log increments (from 2 to 20% contrast) compared to only one contrast in Experiment 1.
The procedure was similar to the procedure in Experiment 1. However, the angle of the Gabor patches was now determined on the performance on the practice trials for every participant. The angle could either be 1, 2, 3, 4, 5, or 6 degrees (all presented 16 times). The smallest angle was chosen for which performance >.8. If performance did not reach.8 a 6 degree tilt was used in the remainder of the experiment. Participants performed 96 practice trials and a total of 1440 experimental trials. Half of the experimental trials contained tilted and the other half untilted targets (720 trials each). Participants could take a break after every 72 trials. The experiment lasted about 80 minutes.
The effects of cue type on tilt detection (as measured by
The contrast of the target exhibited a strong effect on performance,
In our analysis with Block and Cue type as factors, there was again no main effect of cue type on tilt detection,
The cues did not alter participants’ response criterion,
Also in our analysis with blocks as factor there was no main effect of cue type on response bias,
Novelty has long been known to have distracting effects by evoking an orienting response towards new things in the environment
In the present study, the effect of novel stimuli on visual perception was investigated in two experiments. If novelty would enhance visual perception by recruiting attentional resources, one would expect to find higher sensitivity on the trials in which the cue was novel. The first experiment revealed better tilt detection after novel cues for different degrees of tilt. These effects tended to grow stronger over experimental blocks (statistical trend). Possibly the familiar cue became more familiar over repeated presentations, making the novels stand out more and increasing their effects on perception. We did not find a shift of the response criterion, suggesting that the sensitivity was enhanced for novel cues, rather than that participants changed their response behavior. In the second experiment the effect of novel compared to familiar cues on contrast sensitivity was investigated. Similarly, it was found that novel cues facilitated visual perception in later experimental blocks; that is, contrast sensitivity as measured by
We hypothesize that the enhancing effects of novelty are mediated by transient attention
As proposed by Weichselgartner and Sperling
There is an alternative explanation of our findings. Possibly, repeated presentation of the familiar cues reduced attentional orienting due to low-level adaptation. To test this alternative, we reasoned that there should be some release from adaptation if a familiar follows a novel cue trial, as compared to familiar cue trial. In both experiments, we found no evidence for such an effect. However, it might still be that the better perception of the target after novel cues relative to familiar cues might not reflect an enhancement after a novel cue, but a normal attentional response to the novel cues compared to a reduced attentional response to the repeated familiars. This explanation would be the other side of the coin relative to the one given above: The significance of novelty to the brain might be the absence of habituation. It is difficult to disentangle these effects, since both interpretations would lead to almost the same pattern of results – better performance after novel compared to familiar cues. However, a reduced attentional orienting account would predict reduced performance for the familiar cues over experimental blocks when familiarity increased. In contrast, in both Experiment 1 and 2 such an effect of the familiar cue over experimental blocks was not significant, whereas the novelty effect was strengthened over experimental blocks in Experiment 2. These findings suggest that increased familiarity did not affect performance as strongly as novelty did.
One way in which cue novelty could affect performance on the task is by suppressing eye movements. If participants do not fixate at the center of the screen throughout the trial, their performance at the tilt detection task would be affected. Novelty could help participants fixate through increased visual stimulation. However, such an explanation would be hard to maintain given the timing in our experiments: the interval from the onset of the cue to the offset of the target stimulus was only 170 ms, shorter than all but the fastest express saccades.
Emotional stimuli already have already been shown to exhibit beneficiary effects on early visual perception, but this is the first time that novelty is shown to exhibit such effects. It is known that novel stimuli activate similar pathways in the brain as emotional stimuli. Specifically the amygdala has been associated with novelty detection and processing (Blackford, Buckholtz, Avery and Zald, 2010). Possibly, via these emotional pathways in the brain, novelty enhanced tilt detection in the present study. A recent functional magnetic resonance imaging (fMRI) study reported that emotional stimuli enhance perceptual processing via amygdala back-projection to the inferior temporal cortex