In our daily life, anticipation or expectation might help people avoid harm or increase performance. In cognitive neuroscience, anticipation or preparation has been found to affect brain activity in a few seconds before an upcoming event (Brunia, van Boxtel, &
Böcker, 2012). The earliest research on neural correlates of expectation can be tracked back to 1964. Walter, Cooper, Aldridge, McCallum and Winter (1964) presented a series of irregular or regular pairs of stimuli. In the irregular presentation, stimuli were
44
presented at an interval of 3-10 seconds while regular pairs of stimuli were presented at a regular interval of 1 or 1.5 second(s). The first stimulus (S1) was a click and the second stimulus (S2) was a flash. Participants needed to press a button when they saw the flash. A slow negative deflection was observed over frontocentral scalp sites and reached its maximum around 20 V just before S2 of the regular pairs. Walter et al.’s findings indicate that the second stimulus in the regular pairs is the key point. The Contingent Negative Variation (CNV) disappeared when S2 was not presented and restored when the second stimulus was re-presented. Therefore, the CNV has been thought to reflect preparation for responses to S2. However, Walter et al. also indicated that the CNV could be observed as long as participants kept being attentive and pressed buttons immediately. This was criticised because the CNV may not only reflect preparation for the responses but also attention to S1 (Järvilehto & Fruhstorfer, 1970).
There were at least two components of the observed CNV. One was an earlier frontally distributed waveform that was related to S1 and the other was a late centrally distributed waveform just before S2. The latter one was called a ‘readiness potential’. The later potential reflected a preparation for responses to S2. Correlations between RTs of responses to S2 and magnitude of the later waveforms seem to support this idea. The more negative waveforms before S2 were, the faster RTs to S2 were (Brunia &
Vingerhoets, 1980; Hillyard, 1969). More recently, CNV is thought to be an index of anticipation timing, i.e. temporal expectation (Nobre, Correa, & Coull, 2007). In the studies investigating temporal expectation, the temporal information of the fore period was manipulated by changing S1. If S1 indicates the fore period is more predictable and short before S2 onset, not only the RTs to S2 are faster but also amplitudes of CNV are larger before S2 (Miniussi, Wilding, Coull, & Nobre, 1999; Zanto et al., 2011). Macar and Vidal (2003) asked participants to compare durations between S1 and S2 with the target duration 2 s. Participants had to press a button to indicate whether the duration
45
was same (yes) or different (no) compared with 2 s when S2 was presented. Magnitude of CNV reached their maximum at the estimated target time point even when the duration was longer than 2 s. This suggests that the peak of CNV amplitudes might not only reflect preparation for actions but also internal evaluation of temporal information.
The late CNV amplitudes are also affected by an interaction between motivation and attentional resources. A recent study by Schevernels, Krebs, Santens, Woldorff and Boehler (2014) found that when S1 indicated that the task of S2 was more difficult and monetary reward was given, CNV amplitudes were larger than when S1 indicated the task of S2 was easier. No differences in CNV amplitudes between difficult and easy tasks in no reward trials were found, as there might be no motivation to spend more effort on preparing for a more difficult task. RTs to S2 were faster when the differences of CNV amplitudes between difficult and easy tasks were larger but only in reward trials. These findings suggest that CNV amplitudes reflect both motor preparation and attention to an upcoming event. As an index of preparatory processes, CNV can be influenced by reward in terms of allocating preparatory effort.
It has been suggested that the later component of the CNV reflects a ‘readiness potential’ that is related to motor preparation (Järvilehto & Fruhstorfer, 1970).
Differently from CNV which is a component between two stimuli, the ‘readiness potential’ is a component that reflects preparation before a movement whether a stimulus is or is not presented. It is also known as the Bereitschaftspotential (BP), which is termed by Kornhuber and Deecke (1965). BP is also a slow negative shift over central scalp sites, starts from 2 s and shows an increase in negativity very shortly before a movement is made (for a review, see Shibasaki & Hallett, 2006). BP is observed by asking participants to do voluntary movements for example pressing buttons at self-paced rates. The source of BP is suggested to be motor cortex and lateralised depending on response hands (Cunnington, Windischberger, Deecke, & Moser, 2003; Shibasaki &
46
Hallett, 2006). There is another component related to anticipation or preparation but more reflecting attention rather than action preparation. It is termed as Stimulus-Preceding Negativity (SPN) and was observed in a time estimation task (Damen &
Brunia, 1988). Damen and Brunia (1988) asked participants to press buttons once per 20 s. A stimulus was presented 2 s after button presses as a feedback of the time that participants made responses. This distinguished the component of motor preparation from anticipation to a stimulus. The BP was found largest contralateral to the movement side before button pressing while another negative shift (SPN) was found largest over the right hemisphere before the feedback stimulus regardless of the movement side.
They suggested SPN reflects a different process from BP due to its scalp distribution.
SPN may be the component of attentional preparation for upcoming stimulus. Overall, preparation on actions or upcoming events can be observed by slow negative EEG waveforms preceding actions or stimulus. SPN reflects purely non-movement-related attentional anticipation. BP is a preparatory component on movement. CNV might be an ensemble of BP and SPN that include both anticipation to action and stimulus.
The early studies of CNV mainly used a warning stimulus as S1 to indicate the other stimulus is going to be presented. More recently, an S1 is not merely a warning stimulus but contains information about fore periods or locations of S2. The anticipatory process is from ‘expecting an event’ to ‘knowing what to expect’. Reaction times are significantly faster if a valid cue indicates orientations or temporal positions of a stimulus (Nobre et al., 2007; Posner, 1980). Information of a valid cue might play a role of top-down control in anticipation of an event. It was found that a visual cue that indicates whether colour or motion of an upcoming stimulus needs to be attended could activate brain areas that are responsible for colour and motion processing in the pre-stimulus period (Shibata et al., 2008). It was also suggested that when different category of objects are instructed to be attended, corresponding brain areas are activated even
47
before stimulus onset (Driver & Frith, 2000). In a working memory task, Bollinger, Rubens, Zanto and Gazzaley (2010) used informative cues to indicate categories of upcoming pictures (faces or scenes) compared with non-informative cues. They found that working memory performance was significantly better after informative cues than non-informative cues. In addition, the Fusiform Face Area (FFA) was activated and showed more functional connectivity with only the prefrontal cortex when the category of an upcoming picture was known to be a face. A recent study investigated how anticipation optimises brain activity in working memory updating (Yu, FitzGerald, &
Friston, 2013). The dopaminergic midbrain and striatum was activated when high probability of working memory updating was cued. In another working memory study, cue-related activity could predict subsequent working memory performance (Murray, Nobre, & Stokes, 2011). These findings suggest that if pre-stimulus cues contain valid information about properties of upcoming events, both pre-stimulus brain activity and behaviour on upcoming events could be optimised. The valid information about upcoming events might enable imagining processes to optimise brain activity (Driver &
Frith, 2000). If any information about the properties of a target is given beforehand, for example, the category of an upcoming picture will be a face, and then imagining a template of a face could activate the FFA.
Unfortunately, in our daily life, sometimes there is no cue to guide our behaviour or brain. Things are happening by series. In this case, it was suggested by Barton, Kuzin, Polli and Manoach (2006) that knowledge about upcoming events in a series could be obtained by maintaining and updating information of rules of the series, which involves executive processes especially working memory. In addition, attention can tune the precision of perceptual expectation. Using Multivariate Pattern Analysis (MVPA), Jiang, Summerfield and Egner (2013) found that attention helped to differentiate neural patterns of expected stimuli from unexpected stimuli. Additionally,
48
using MVPA, Haynes et al. (2007) showed that activity in the Medial PreFrontal Cortex (MPFC) could predict participants’ intentions as to which task they were going to do.
Most importantly, in that study, participants were asked to choose one of two tasks freely and hold their decision in their minds for variable delays before stimulus onset.
MPFC, rather than lateral PFC, patterns predicted later task selection. The MPFC is related to goal-directed information selection processes and this predicted participants’
later intentions most accurately. This might be due to participants choosing a task freely, which is different from that a task goal is decided by the experimenter. This finding suggests that preparatory brain activity for a specific task may be under participants’
strategic control.
Some brain states can also help later perception or responses if no information is available before the stimulus. For example, a stimulus presented after decreased phase coupling in alpha frequency band is more likely to be perceived (Hanslmayr et al., 2007;
Mathewson, Gratton, Fabiani, Beck, & Ro, 2009). Likewise, power increases in pre-stimulus alpha activity over occipital regions can impair later visual discrimination (van Dijk, Schoffelen, Oostenveld, & Jensen, 2008). In a go/ no-go paradigm, larger occipital alpha power before stimulus onset led to more errors in the task (Mazaheri, Nieuwenhuis, van Dijk, & Jensen, 2009). The findings indicate that increases in pre-stimulus alpha activity reduce accuracy in perception. In contrast, performance is better when less pre-stimulus alpha activity is available. Pre-stimulus alpha activity might gate visual stream to play an inhibitory role in perceiving an upcoming visual event (Jokisch
& Jensen, 2007; Mazaheri et al., 2009).