Working memory can be approached from two points of view: the impact of a memory load on the processing of information or the impact of processing activities on memory
performance. Baddeley and Hitch (1974) had started the working memory research by investigating the influence of a memory load on processing activities, while further research has rather concentrated on the maintenance of information in the presence of interfering processing activities. Focusing on maintenance, the most recurrent research questions concern the structure of working memory, its resources and its capacity. It is quite evident that these three issues are largely intertwined and advances in the understanding on one of them have direct consequences for the others. Although research has exponentially progressed over the years and many propositions have been made to elucidate the functioning of working memory, as many questions remain unanswered.
The goal of the current research project frames within a very actual research trend within working memory investigations: the binding problem. This specific subpart of working memory research focusses on the encoding and maintenance within working memory of complex information, i.e. information composed of multiple features like for example a green circle. Earlier working memory research has especially concentrated on the maintenance of single features such as letters, shapes or locations. This was a necessary step to get
information about the basics of working memory. However, the world around us isn’t made up of simple letters, shapes and locations. More complex problems have hence become the topic of investigation, like the maintenance of associated information. What does it take to remember that a particular circle was green and not blue? Several working memory models have proposed a certain structure or framework for the working memory system. However the representation of complex information within this structure is often poorly defined. The same is true for its capacity. Capacity estimates exist mainly for the maintenance of simple letters, numbers, shapes or locations, but what about the capacity of working memory when it has to deal with complex information. Can only a fixed number of features be maintained? Or does one object composed of several features counts only as much as one simple feature? Next, the multi-component model of Baddeley and Hitch (1974) has suggested domain-specific
resources to exist to maintain information belonging to different domains. For example verbal resources are assumed to be responsible to maintain verbal information, while shapes are rather maintained by a visuo-spatial resource. But what resources come into play when
maintaining complex information that is built up of information belonging to different domains?
The goal of this thesis is to increase our understanding of working memory when dealing with complex information. Our research on the maintenance of complex information has concentrated on the same questions that were asked when dealing with single features, namely 1) how is this information represented within the working memory structure 2) what resources does its maintenance depend on 3) what is its capacity limit. As a typical instance of complex information, we will focus in the present research project on the maintenance of cross-domain information that is composed of a verbal and a spatial component. Although this concerns just one instance of complex information limiting as such general conclusions, it will become clear while reading through the introduction that this particular combination of verbal and spatial information is at the moment the most appropriate to reach conclusions that are generalizable.
Within this thesis we sought first of all to find answers to our questions on the maintenance of cross-domain information, as well as to broaden our existing knowledge on the bases of working memory. As stated before, many questions on the foundations of working memory remain to be elaborated. A narrow focus on these issues has not resulted in general consensus among researchers. By tackling more complex problems, we adapt to a broader point of view on working memory which might possibly shed new light on the more basic questions. Though the research questions as addressed here are quite theoretical, practical implications are manifold and of all sorts. One might for example wonder how, when and where traffic signs should best be presented in order to avoid confusion leading to accidents on the road. Traffic signs are typical example of our dependence on working memory as the information presented on it has to be maintained until the moment we need it, while in the meantime stimuli from everywhere around us keep coming and need to be adequately processed. Should we present all information on one and the same traffic sign or spread them over multiple traffic signs presented every now and then as we advance our trajectory? The answer on this kind of questions depends thus on our more theoretical
questions on the capacity limits of working memory and more precisely on the capacity limits for complex information. Another example of the utility of this kind of studies is for example within a learning context. How can we increase the opportunity for information to be
maintained? The goal of learning is to get as much information maintained as possible, within a minimum of time and with a minimum of effort. Knowing how the working memory system
operates could be a real help to obtain this objective. (Who wouldn’t be grateful if he or she could automatically and unlimitedly maintain all the information he or she wanted to?)
The three main research goals addressed within this thesis: 1) How is cross-domain information represented within the working memory structure? 2) What resources does the maintenance of cross-domain information depend on? 3) What is the capacity limit of
working memory for cross-domain information? will first be addressed in light of the existing literature in order to elucidate the need of the present research project. To do so, we will mainly rely on three working memory theories: the multi-component model (e.g., Baddeley, Allen, & Hitch, 2011; Baddeley & Hitch, 1974), Cowan’s embedded process model (e.g., Cowan, 1988, 2005) and the time-based resource sharing model (e.g., Barrouillet, Bernardin,
& Camos, 2004; Barrouillet & Camos, 2015). These three models are all three well known theories within the field of working memory and have elaborated considerations concerning the representations, the resources and the capacity limits of working memory. Next to the fact that they represent “great theories” of working memory, they are complementary for the goals of this thesis as each one of them has particularly elaborated one of the three research topics.
While the multi-component model is very detailed in terms of its architecture, the time-based resource sharing model is more detailed in terms of the application of the resources and the embedded process model has especially elaborated the capacity limits of working memory.
In the first chapter of this thesis we will discuss the representation of information within the working memory system. For each of the three to-be-presented working memory models, we will start by introducing the basics of the model and how single feature
information is represented. We will then continue by discussing how the question of the representation of associated features has been tackled. In the second chapter we will discuss the resources necessary to maintain working memory representations. As in the first chapter, we will start by discussing the resources used to maintain single features within working memory and subsequently discuss the resources for the maintenance of associated features.
The third chapter concerns working memory capacity. This chapter follows the same structure as the previous ones, attending first to the capacity limits for the maintenance of single features within working memory, followed by a discussion on the capacity limits for
associated features. In the fourth chapter we will explain the rationale and hypothesis for the upcoming experiments, based on the observations discussed in the first three chapters. The experimental plan will also briefly be introduced. The fifth, sixth and seventh chapter will exhaustively explain the methodology of the executed experiments and the obtained results concerning the working memory capacity, resources and representations for the maintenance of cross-domain associations respectively. A brief discussion will be presented after each experiment and at the end of each of the three chapters. In the eighth chapter, we will discuss the totality of obtained results and relate them to the observations on the maintenance of associated features discussed in the first three chapters. A final conclusion regrouping the most important findings of this research project will be presented at the end.