Studies employing a masked priming paradigm have observed that subliminal stimuli could be processed on a semantic level. Semantic categorization of a consciously perceived stimulus could be facilitated by the presentation of a former stimulus that was not consciously perceived (subliminal). This facilitation is usually measured as the reaction time employed to respond to a task. When this facilitation occurs between congruent pairs of stimuli (from the same category) and does not occur between incongruent pairs (from different categories) is called congruency priming effect. Two principal factors modulate the subliminal priming effect, the semantic similarity between stimuli and the SOA. Semantic similarity refers to the similarity in meaning or overlap of features between two words. The greater the similarity, the bigger the facilitation. Another factor that modulates priming effect is the interval between the onset of the first stimulus and the onset of the second stimulus of the par (SOA: stimulus onset asynchrony), showing that the bigger the SOA, the lower priming effect. Despite the fact that these two phenomena have been extensively studied, it has not been studied the influence of semantic strength has not been studied together with SOA duration. The goal of this study is to observe if semantic relatedness (strongly and weakly related pairs) could modulate congruency priming effect duration. To evaluate this, both semantic relatedness (strong and weak) and SOA were manipulated in a subliminal semantic priming task.

The etymological meaning of the word subliminal comes from the Latin Sub (below) and Limen (threshold) and refers to an information processing below the threshold of consciousness . A growing body of behavioral and electrophysiological data shows that automatic cognitive processes, such as language processing, can be performed under certain circumstances without conscious resources (Dehaene, Changeux, Naccache, Sackur & Sergent, 2006.) The term subliminal perception refers to the inability on the part of a subject to report having perceived the apparition of a stimulus, and nevertheless notice that the behavior of the subject (i.e. motor behavior, decisions, thoughts, feelings, or even learning) has been influenced by this unconscious stimulus. This inability is due to an insufficient force of the stimulus to achieve an attentional response, generally due to temporary limitations (i.e. very rapid presentation of the stimulus).
The study of the processing of subliminal stimuli at the semantic level can be approached through the masked priming paradigm. In this paradigm, a series of stimuli are presented to participants, who are usually asked to categorize them (e.g., deciding if the stimulus is an animal or an object). The stimulus to which the participant decides upon is completely apprehensible by conscious perception and it is called target. In the masked prime paradigm, target stimuli are preceded by a stimulus called prime. To achieve subliminal status, this prime stimulus must be presented for a brief period (usually, less than 50 ms) and visually masked (Kouider & Dehaene, 2007). The masking usually implies the use of two stimuli which are given before (forward mask) and/or after (backward mask) the prime stimulus. These masks are configured to prevent the prime stimulus from being consciously perceived, but still be able to influence the behavior of the participants (Figure 1). In a semantic categorization task, this is measured by a difference in the reaction times (“RTs”) between the response given when the prime is congruent (i.e. both stimuli sharing the same category) with the target and when the prime is incongruent (different categories) with it. This effect is called masked congruency priming effect.

Figure 1. Representation of a single trial in a subliminal priming task.

The effect of masked priming can be modulated by various factors, but three factors stand out above the others. The automatic processing of the semantic priming seems to be associated with two characteristics of the prime: 1) The strength of association between the prime and the target, and 2) The semantic similarity between prime and target. The latter would be the similarity in meaning or the overlap of features between two words (i.e., cat and tiger are strongly related given that they share attributes such as fur, claws, being felines, etc.) (McRae and Boisvert, 1998). The strength of association refers to the likelihood that a stimulus (i.e., glass) will bring the mind to a second stimulus (i.e., water), despite not sharing attributes. Since these two characteristics (semantic similarity and strength of association) often occur together, the term semantic relationship is used in a broad sense to refer to both (Ortells et al., 2016). The third important modulator of the masked priming effect is the interval between the onset of the prime and the onset of the target stimulus (SOA: stimulus onset asynchrony). The subliminal priming effect at the semantic level tends to be very short-lived compared to the conscious (supraliminal) priming effect, in which the subjects are aware of the existence of the prime. The subliminal priming decays rapidly, suggesting that the mental representation of the subliminal stimulus vanishes as time passes (Dehaene and Naccache, 2001). This is consistent with the findings showing that the bigger the SOA, the lower the priming effect reported (Greenwald et al., 1996)

Although SOA has proven to be an important factor when studying priming phenomena, there is no consensus among researchers about the duration of the prime stimulus in consciousness. Greenwald et al. (1996) argue that to obtain a subliminal priming effect at the semantic level, the SOA should be very short (i.e. 67 ms.) since this effect decays for SOAs bigger than 100 ms. However, other studies observed evidence of subliminal semantic priming effect with bigger times. Ortells et al. (2016) performed an experiment using a 200 ms. SOA in which they manipulated the semantic similarity between pairs of words. The subjects were exposed to three possible pairs of stimuli, unrelated stimuli (dog - head), weakly related (dog - head) and strongly related (dog - cow). The authors observed a subliminal semantic priming effect only for strongly related pairs. In a replication of Ortell et al (2016), Bruno, Diaz Rivera, Embon & Iorio (2016) compared the effect of priming on a masked priming task using both pairs of words and a standardized set of images taken from Rossion y Pourtois, (2004). Using a similar methodology to Ortells et al (2016), they provided three possible categories of pairs (not related, weakly related and high related) with a SOA of 67 ms. The authors reported a priming effect for high related images, but not for words. Authors suggested that the absence of a priming effect for words could be due to differences in the semantic similarity for the pair words employed, occasioned by the translation from English to Spanish. The authors propose the use of images as a more reliable and standardized way to study subliminal semantic priming.

Despite that both SOA and semantic similarity influence have been extensively studied, it has not been studied the influence of semantic similarity and SOA together. The main objectives of this study were two. First, replicate results of former studies. In other words, observe whether the modulation of semantic similarity on a masked priming task leads to a difference in RT. Second, determine a SOA limit for subliminal priming using pairs of images related with different semantic strength. In other words, evaluate if the priming effect duration depends on the semantic similarity of stimuli and to what extent. These results will contribute to expanding current knowledge about subliminal semantic processing. At the same time, they will offer new evidence about the role of consciousness in the processing of stimuli.

For a detailed descripton of the result consult the final report. Or you can have access to a PDF file version of the poster presented at the conference.