What determines the direction of subliminal priming

Masked stimuli (primes) can affect the preparation of a motor response to subsequently presented target stimuli. Reactions to the target can be facilitated (straight priming) or inhibited (inverse priming) when preceded by a compatible prime (calling for the same response) and also when preceded by an incompatible prime. Several hypotheses are currently under debate. These are the self-inhibition (SI) hypothesis, the object-updating (OU) hypothesis, and mask-triggered inhibition (MTI) hypothesis. All assume that the initial activation of the motor response is elicited by the prime according to its identity. This activation inevitably leads to straight priming in some cases and the mechanisms involved are undisputed. The hypotheses differ, however, as to why inverse priming occurs. The self-inhibition (SI) hypothesis assumes that the motor activation elicited by a prime is automatically followed by an inhibition phase, leading to inverse priming if three conditions are fulfilled: perceptual evidence for the prime has to be sufficiently strong, it has to be immediately removed by the mask, and the delay between the prime and target has to be long enough for inhibition to become effective. The object-updating (OU) hypothesis assumes that inverse priming is triggered by the mask, provided that it contains features calling for the alternative response (i.e. the one contrasting with the response induced by the prime). The MTI hypothesis assumes that the inhibitory phase is triggered by each successive stimulus which does not support the perceptual hypothesis provided by the prime. Based mostly on our own experiments, we argue that (1) attempts to manipulate the three factors required by the SI hypothesis imply changes of other variables and that (2) indeed, other variables seem to affect priming: prime-mask perceptual interaction and temporal position of the mask. These observations are in favor of the MTI hypothesis. A limiting factor for all three hypotheses is that inverse priming is larger for arrows than for other shapes, making it doubtful as to what extent the majority of studies on inverse priming, due to their use of arrows, can be generalized to other stimuli.


Subliminal priming
R������� o� u��o����ou� ��flu����� o� b�����o� ��� b��� attracting much interest in recent years because these kinds of studies give some insight into what conscious awareness is for. Indeed, by comparing what the human brain can do without bothering for conscious awareness to those situations where conscious mediation is needed, we can try to infer the role of the latter process.
To make stimuli unseen and still effective on be� havior, usually backward masking is used in which a These are the self-inhibition (SI) hypothesis, the object-updating (OU) hypothesis, and masktriggered inhibition (MTI) hypothesis. All assume that the initial activation of the motor response is elicited by the prime according to its identity. This activation inevitably leads to straight priming in some cases and the mechanisms involved are undisputed. The hypotheses differ, however, as to why inverse priming occurs. The self-inhibition (SI) hypothesis assumes that the motor activation elicited by a prime is automatically followed by an inhibition phase, leading to inverse http://www.ac-psych.org Piotr Jaśkowski and Rolf Verleger subsequent stimulus (the mask) is able to reduce vis� ibility of the preceding stimulus (the prime). With the appropriate timing and spatial arrangement of masks and primes, this technique works very effectively with a ��d� ����� o� �t��u��. A �u����t�� ��d�������d ���ot��� sis assumes that the mask disrupts the reentry process (an iterative loop comparing sensory input with stored representations), which is thought to be necessary for creating a vivid percept (Di Lollo & Enns, 2000).
They measured simple reaction times (RT) to presenta� tions of a square. With onset asynchronies ranging from 0 to 75 ms, two other squares were displayed left and right of the original one, masking the priming square by metacontrast (Breitmeyer, 1984) and thereby reducing its perceived brightness to an extent depending on the interval between the prime and mask, with a maximal reduction for the onset asynchrony of 75 ms. Because o� t�� �������o�� ������� d����d���� b�t���� ������ reaction time and brightness (e.g. Bartlett & MacLeod, 1954;����o����� 1985;M���fi��d� 1973), Fehrer and Raab expected the longest RTs for 75 ms. It turned out, however, that RT did not depend on perceived bright� ness at all. Fehrer and Raab suggested that the primes triggered reactions before their perceived brightness was reduced by the mask (Note 1).

Inverse priming
The DPS theory suggests that what is formed consciously in a choice task is an intention. Those stimuli encom� �����d b� t�� ��t��t�o� ��� t��� b� �d��t�fi�d �uto��t�� cally, without mediation of consciousness, and can trigger response activation. Therefore, if some prime, visible or not, is incompatible with the target, the wrong response is initially activated. When the target appears, the ongo� ing motor activity has to be canceled and replaced by the preparation of the alternative response.
Three shapes -a prime, a mask and a target -were �o����ut����� d�������d �t fix�t�o�. T�� ����� ��d t�� t����t ���� doub�� ���o�����d�� �o��t��� to t�� ���t or to the right. Participants had to respond with their left or right hands depending on whether the arrows pointed left or right. The mask was formed from the two target shapes overlaid on one another. The pattern of results was different from that obtained with metacontrast masking: the RTs were shorter and more accurate when the priming and tar� get arrows pointed in different directions (incompatible trials) than when they pointed in the same directions.
Throughout this article we will refer to this phenom� enon as "inverse priming".

SELF-INHIbITION
In the light of the theories at the time, Eimer and Schlaghecken's (1998)

Prime visibility and strength of sensory representation
To account for further results, Schlaghecken and Eimer (2002) had to introduce another assumption, namely that an important factor determining the sign of the priming effect is the visibility of the prime. First of all, inverse priming had never been noted when the prime was left unmasked (Klapp & Hinkley, 2002;Verleger et al., 2004). Moreover, Eimer and Schlaghecken (2002) showed that the priming effect increased from negative
Nevertheless, occurrence of the inhibition phase of the LRP is strictly related to the moment of mask pres� entation. This may be taken to suggest that a more critical factor than occlusion from visibility is just the presentation of a temporally trailing stimulus at the same place or in the nearest vicinity.

Mask structure matters
The size of inverse priming (Eimer & Schlaghecken, 1998)  Eimer and Schlaghecken's (1998) approach was generalized by Klapp and Hinkley (2002) to encom� pass the difference between conscious and uncon� scious processing. These two papers were criticized by Lleras and Enns (2004) and Verleger et al.

What is updated in the scene?
An important question to be asked is what exactly is updated in the scene when a mask appears. priming with masks which were formed from some ar� rows randomly distributed over an area (Fig. 2). What �bout ����� u�d�t��� �� t��� ����? O�� ��� ���u�� that an object is more abstract than just a shape at a  Krawczyk (2005  and V�������� E����� ��d ����o���� (submitted), we concede that inhibition of the primed response does occur. The main difference from Schlaghecken and Eimer's view (Bowman et al., 2006; is that this inhibition is considered to be evoked by the mask rather than being a rigid consequence of prime activation occluded from further ������tu�� ���d����. Mo�� �����fi������ �� ���u�� t��t each immediately following stimulus appearing within the focus of attention, which does not support the perceptual hypothesis concerning the prime's identity (Note 5), will inhibit the ongoing action and thereby activate the alternative response (Note 6). However, the mask produces this inhibition particularly if and insofar as it contains elements similar to the primes:

Results of an experiment by Jaśkowski and Przekoracka-
Perception of any such elements informs the system that activation was premature and should be inhibited.
Thus, we propose that the mask acts as a "false friend" to the processing system. It is a friend by preventing the system from misperceiving the primes as targets. But it is a false friend by presenting features that might be misunderstood by the system as being relevant. As a ������u��d ������t t��� ��t���������� �x��t��� ����o��� activations get inhibited, putting them at a disadvantage for the upcoming response to the target.
T���� fi�d���� ���o �o���d Lleras and Enns (2006)  The second assumption they added was called the repeated location advantage. It refers to the observa� tion that inverse priming is more likely when the prime and target are presented at the same or some nearby location. Lleras and Enns (2005) showed that when http://www.ac-psych.org the prime and target were presented at the same position, inverse priming occurred also for irrelevant masks (consisting of vertical and horizontal lines). In contrast, when only the prime and mask were present� ed at the same position, while the target was displayed aside, inverse priming occurred only for the relevant mask. Lleras and Enns (2006) argued that the visual system considers a spatiotemporally proximal prime and target as two instantiations of a single object which changes/develops in time. �o����t��t ��t� t�� O�+ ���ot������ �����/t����t/ ���� ��������t�� �b�u�t o���t o� fl������� ��d "���t��� similarity", that is, the proximity of the presented ob� jects, made the priming effect more negative.
T�� O� ���ot����� (���� t�� ot��� t�o) d����o�� "by budding": once a problem is encountered, a new assumption is added. At the same time, no clear evi� dence has been provided that enhancement of inverse priming with relevant masks is due to object updating.
Above we argued that the MTI hypothesis provides an alternative explanation.

Results of two experiments by Jaśkowski and Ślósarek (2007) with primes of different shapes. Priming effect [= RT(incompatible) -RT(compatible)
] is plotted as a function of the prime-target interval. The shapes of the primes used are shown near each plot. In the experiment whose results are presented in the left graph, the masks were formed from lines of different orientation and length, randomly dispersed over an area. The mask used in the other experiment was formed from the two primes of a given pair overlaid with one another. Note that overlaying the two pairs of the primes forms the identical mask.
(submitted) obtained inverse priming with arrows only.
They did obtain a differential effect of masks on letter