El modelo de las Páginas Amarillas de Expertos

This section focuses on three well-known phenomena related to the RSVP tech- nique: the Attentional Blink (AB), attentional awakening and the intertrial prim- ing effect. Even though this review will not go very deep into these concepts, due to their relevance in the RSVP literature, it is important for the next sections to give the readers a clear idea of this phenomena in order to highlight the limitations of this technique.

2.4.3.1 The Attentional Blink

The attentional blink is a phenomenon that refers to a target being missed by participants when it is presented within a short time interval after the previous target in the same stream of stimuli [Chun & Potter, 1995; Cinel et al., 2004;

Einh¨auser et al.,2007; Kranczioch et al., 2003].

second target) vary depending on the type of stimuli being presented [Einh¨auser et al., 2007; Most et al., 2007].

A possible explanation of the AB is that after the detection of the first tar- get, some resources will be allocated to the processing of that stimulus, temporaly decreasing the capacity of the (limited) visual system [Di Lollo et al.,2005;Kawa- hara et al., 2006].

In order to avoid the attentional blink, Kruse & Makeig [2007] allowed a maximum of one target in every 51-picture stream. Furthermore, targets were placed within the central part of the image array to avoid crossover effects, since they reported that targets could not be detected in the last two images due to the “shock” produced by the end of sequence [Kruse & Makeig, 2007] (see Section 2.4.3.2).

Many of the studies described in the previous section have the same limitation with regard to the proportion of target vs non-target items, or on the way in which these are placed within a stream, e.g., allowing only one target within a stream of images [Mathan et al., 2006, 2008]. On the other hand, Healy et al.

[2010] constructed a more generic environment in which the proportion of targets vs non-targets was fixed at 10% — the typical target occurrence in the oddball paradigm, and an ideal ratio for eliciting large P300s [Johnson, 1986] —, and their results were not affected by this change. The same ratio was used byCecotti et al. [2012, 2014a, 2015]; Marathe et al. [2015], although occasionally imposing restrictions on the inter-target interval to avoid attentional blinks [Marathe et al.,

2015].

When considering the effects of the AB in experiments and real life applica- tions, the real expected number of targets should be taken into account. In broad-

area search, for example, the occurrence of targets may be very low (1% according to Mathan et al. [2008]), but targets could appear in images that show regions that are near each other. If neighbouring images are presented in close temporal proximity, as one would normally assume, this could produce attentional blinks and subsequent target misses. In this case, shuffling the images could solve the problem of the AB. On the other hand, the occurrence of images considered to be targets will be higher on medical data sets, a fact that should be taken into account in the labs, specially when deciding an appropriate presentation rate, since higher rates will make targets come closer in time.

Although it was recently shown that the AB can be decreased through train- ingChoi et al.[2012], further developments in this area showed that the processing limits that are signalled by the occurrence of the AB cannot be eliminated [Tang et al., 2014], and that the improvements achieved by Choi et al. [2012] may ac- tually be related to the attentional awakening and intertrial priming effects (see below).

Lastly, even if a participant was able to detect the two targets, the refractory period of the P300 should be taken into account. In the case of two targets occurring very closely in time, the evoked P300 component (if at all present) will be of smaller magnitude, and might thus be misclassified by the system.

2.4.3.2 Attentional Awakening

As opposed to studies considered in Section 2.2.2, which looked into the spatial attentional filtering signalled by the N2pc, the RSVP technique is used to study the temporal modality of attentional filtering [Riccio et al., 2013; Shapiro et al.,

processing [Chun & Potter, 1995].

The attentional awakening consists of a gradual increase in detection accu- racy over time during an RSVP burst [Ariga & Yokosawa, 2008; Kranczioch & Dhinakaran, 2013]. This means that targets that are presented at the beginning of a burst are more likely to be missed than targets that occur at later time positions. According to Ariga & Yokosawa [2008], the attentional awakening represents the observer’s modulation of attention to adjust to the presentation rate in preparation for detecting the target in the sequence, since no significant variations of performance in target detection are found after the preparation is complete [Ariga & Yokosawa, 2008]. Kranczioch & Bryant [2011], however, ar- gues that this phenomenon is not related to a variance in the focus of temporal attention, but rather to resource allocation to the task.

2.4.3.3 Intertrial Priming

The phenomenon by which repeating a target template improves the performance and speed of visual search is known as intertrial priming.

Intertrial priming was first proposed by Maljkovic & Nakayama [1994], and this effect has been observed in visual search scenarios regardless of whether the feature that remains constant for the target template is of the “pop out” type (e.g., colour) [Maljkovic & Nakayama, 1994; Yashar & Lamy, 2010a], re- quires a conjuctive search [Kristj´ansson & Driver, 2008], or remains in the same spatial [Maljkovic & Nakayama,1996] or temporal location [Kristj´ansson & Cam- pana, 2010; Kristj´ansson et al., 2010; Yashar & Lamy, 2010b].

Thus, this phenomenon is not only present in the RSVP paradigm, but might be related to the increased salience of primed features, thus facilitating the allo-

cation of attention to the processing of those features. Neurophysiology and neu- ropsychology studies have shown that this effect might be originated in the neural mechanisms that underlie visual search, including the attentional systems and regions in visual areas [Kristj´ansson & Campana, 2010; Kristj´ansson & Driver,

2008].