Research Article: Redundant sensory information does not enhance sequence learning in the serial reaction time task

Date Published: March 1, 2012

Publisher: University of Finance and Management in Warsaw

Author(s): Elger L. Abrahamse, Rob H. J. van der Lubbe, Willem B. Verwey, Izabela Szumska, Piotr Jaśkowski.


In daily life we encounter multiple sources of sensory information at any given
moment. Unknown is whether such sensory redundancy in some way affects implicit
learning of a sequence of events. In the current paper we explored this issue in
a serial reaction time task. Our results indicate that redundant sensory
information does not enhance sequence learning when all sensory information is
presented at the same location (responding to the position and/or color of the
stimuli; Experiment 1), even when the distinct sensory sources provide more or
less similar baseline response latencies (responding to the shape and/or color
of the stimuli; Experiment 2). These findings support the claim that sequence
learning does not (necessarily) benefit from sensory redundancy. Moreover,
transfer was observed between various sets of stimuli, indicating that learning
was predominantly response-based.

Partial Text

Implicit learning refers to the phenomenon that people are able to
acquire skilled behavior or structured knowledge about their environment in a
seemingly automatic and unconscious fashion. Over the last decades, the serial
reaction time (SRT) task has become a highly productive tool in the investigation of
implicit learning (e.g., Nissen & Bullemer,
1987; for overviews, see Abrahamse,
Jiménez, Verwey, & Clegg, 2010; Clegg, DiGirolamo, & Keele, 1998; Keele, Ivry, Mayr, Hazeltine, & Heuer, 2003). In this task,
participants are required to respond fast and accurately to a particular feature of
successively presented stimuli on a screen – here referred to as response
cues1. Unknown to the
participants, the order of presentation of these response cues is pre-structured,
thereby allowing learning of the structure across training (i.e., sequence
learning). To differentiate sequence learning from general practice
effects, a block of (pseudo-) randomly selected response cues is inserted near the
end of the practice phase. The cost in reaction time (RT) and/or accuracy of this
random block relative to its surrounding sequence blocks is commonly used as an
index for sequence learning. Notably, sequence learning (as indicated by performance
measures) and sequence awareness (i.e., explicit knowledge about the precise
regularity) often do not develop at the same rate during training: Relatively small
increases in awareness are accompanied by substantial increases in response speed
and accuracy. Learning is therefore said to be (partly) implicit.

The aim of Experiment 1 was to explore the effect of congruent, spatial-temporally
coinciding response cues on sequence learning in the SRT task. This was achieved by
using position and color cues; that is, each response was mapped exclusively onto a
stimulus with a specific color that appears at a specific position, so that the
correct response is simultaneously signaled both through the position and the color
of the stimulus. This design has been employed already in a set of studies by
Robertson and colleagues (Robertson &
Pascual-Leone, 2001; Robertson, Tormos,
Maeda, & Pascual-Leone, 2001), who reported better sequence learning
in the combined position and color cue condition than in either of the single cue
conditions. However, the conclusiveness of their findings is unclear after a
detailed look at these studies.

In Experiment 2, we opted for using shape and color features of stimuli, as these are
both arbitrarily mapped onto responses and thus were expected to produce more or
less comparable baseline response latencies. Indeed, in a small within-subject pilot
study on random sequences of stimuli, the shape and color features provided highly
similar baseline RTs. If there exist different stimulus-based (in this case for both
shape and color) learning systems in which sequence-specific processing gains
develop with practice, than it would be predicted that the sequence learning effect
is larger for the condition with combined shape and color response cues than for
either single response cue conditions. In addition to this change of response cues,
Experiment 2 also employed a different transfer phase than Experiment 1. In
Experiment 1we assessed transfer across all response cue conditions, and the
motivation for this was to provide a significant comparison of sequence learning
between different training groups while circumventing the problem of different
baseline response latencies. However, with the pair of response cues in Experiment 2
this was no longer necessary (i.e., providing comparable baseline response latencies
during the training phase was the whole purpose of Experiment 2). In Experiment 2 we
assessed transfer to a cue condition that was new for all participants, namely
responding to position response cues, in order to determine if purely
response-related learning developed, and in order to compare the amount of purely
response-related learning across the different training groups. The rationale is
that testing in a new response cue condition would allow transfer only of purely
response-related learning (e.g., response location learning) and not of sequence
learning that is specific to the response cues from the training phase (e.g., shape-
or color-related sequence learning). This transfer method to explore the nature of
sequence learning has been used before in various studies (e.g., Abrahamse et al., 2008; Cohen, Ivry, & Keele, 1990; Keele, Jennings, Jones, Caulton, & Cohen, 1995; Willingham, 1999; Willingham et al., 2000).

The role of sensory information in sequence learning is one of the major issues of
debate within the SRT literature (Abrahamse et al.,
2010; Clegg et al., 1998). The
current paper contributed to this issue by exploring potential sequence learning
benefits from the availability of redundant sensory information. From the notion
that sensory information plays a significant role in sequence learning (it has been
shown that sequence learning can be based on sensory information; e.g., Clegg, 2005; Remillard, 2003), it may be predicted that sequence learning can benefit
from the availability of redundant sensory information, at least under some
conditions. However, in a study by Abrahamse et al. (2009) no sequence learning benefits were observed from adding congruent
(i.e., redundant) tactile response cues to an otherwise standard SRT task with
visual position response cues. Abrahamse et al. (2009) acknowledged that, on the base of their results, the strong claim
that sequence learning is typically unaffected by redundant sensory information
would be premature. The current study aimed at further exploring this issue in two
experiments. Experiment 1 showed that no additional sequence learning benefits are
observed when redundant position and color response cues are presented at the same
location. Experiment 2 additionally showed that response cues with similar baseline
response latencies also leave the magnitude of sequence learning unaffected. From
the set of experiments that are reported here and in the study of Abrahamse et al.
(2009), then, we believe that there is by
now substantial justification for the claim that sequence learning in the SRT task
does not benefit from redundant sensory information, at least at the current level
of practice. We will now discuss our findings in relation to the representational
nature of sequence learning, both with respect to the informational content that
underlies sequence representations, and with respect to the implicit-explicit