Mood and Creativity over Time in a Bipolar Participant

Depression and creativity do not go hand in hand—be this due to the characteristic depletion of resources in depression (Rude, Hertel, Jarrold, Covich, & Hedlund, 1999), or perhaps due to a focus on convergent rather than divergent thinking in such a state (Akbari Chermahini & Hommel, 2012). Indeed, the state of depression seems to be negatively related to one's engagement with the creative process (To, Fisher, Ashkanasy, & Rowe, 2012). As Verhaeghen, Joormann and Khan (2005, p. 226) have noted, “… there is anecdotal evidence concerning creative productivity in individuals with bipolar disorder that suggests that the depressive episode decreases productivity rather than increases it.” In contrast, as Goodwin and Jamison (2007, p. 397) suggest, “there is some evidence that expansiveness of thought and grandiosity of both mood and thought—common features of mild mania—can result in an increased fluency and frequency of ideas that is highly conducive to creative achievement.” Furthermore, according to reports of creative individuals, especially poets, the state in the creative periods greatly resembles the state in manic periods (Jamison, 1993) in restlessness, openness, excitement, grouchiness, grandiosity, and associativeness. Nevertheless, it is when depression alternates with mania, as in a bipolar affective disorder, that creativity is enhanced (Johnson et al., 2012; Richards & Kinney, 1990). While the birth of new ideas usually occurs during the manic phase, the refinement of these ideas can occur during the periods of melancholy. Negative emotions can, in some cases, foster deep observation and perseverance, both of which assist in the processing of ideas, and lead to increased creativity (De Dreu, Baas, & Nijstad, 2008; Kaufman & Baer, 2002). Conceivably, then, it is the shift in mood within affective space that is conducive to creativity (Desseilles, Chang, Piguet, Bertschy, & Dayer, 2012). Indeed, as Carson (2011, pp. 145–146) has argued, “…research on creativity and mood disorders suggests…shifts in mental states associated with mood may facilitate creativity.”

Given a consensus regarding the two-dimensional nature of affective space (Russell, 1980; Stanley & Meyer, 2009), it is therefore valid to question whether this labile relationship between mood and creativity is dependent on one's current position in that space. If we label the two primary dimensions of this space as Pleasure (P) and Arousal (A), then it would seem that P+A+ enhances creativity, P−A+ diminishes creativity, while P+A− and P−A− are not related to creativity (Baas, De Dreu, & Nijstad, 2008; De Dreu et al., 2008). That, in any event, is what cross-sectional studies of this mood-creativity relationship indicate.

The literature on this mood-creativity relationship has three major interfaces. One concerns the influence of induced mood on creativity (e.g., Ashby, Isen, & Turken, 1999; Fodor, 1999). A second interface addresses the effect of induced speed of thought on both mood and creativity (e.g., Pronin, 2013). The third, of present relevance, concerns the relationship between pathology, and in particular affective disorder, and creativity (e.g., Richards & Kinney, 1990; Taylor, Fletcher, & Lobban, 2015). What, however, do we know about the relationship between mood and creativity in what Schubert et al. (2012) have recently termed “life as it is lived”? The single-subject research design can greatly benefit our understanding of such relationships, as one systematically traces the covariation of mood and creativity over time. One can thereby generate concomitant time series of assessments (West & Hepworth, 1991). In particular, one can answer the question recently posed by Kaufmann and Kaufmann (2014, p. 212): Is there any value in the actual swing of mood for creativity? Conceivably, our participant in the present research might well agree with the comments of comparable bipolar participants in another study, that “… shifting alternate mood states fuelled the creative process by offering them flexibility of thought and contrasting perspectives to generate original ideas” (Taylor et al., 2015, p. 663).

Building on a previous study exploring mood alteration over time (Glicksohn, Gvirtsman, & Offer, 1995–1996), we employ the same data-analytic techniques to investigate the mood-creativity relationship in a single-subject, time-series data set. Of primary interest here is our use of spectral analysis (Larsen, 1990) to investigate fluctuations in mood and creativity, in separate, and in unison. Spectral analysis has been employed to study mood variability, that is to say the “frequency of mood change, not just magnitude of change” (Larsen, 1987, p. 1200) and the existence of particular cyclicity in the data, such as a weekly cycle (Larsen & Kasimatis, 1990; Ram et al., 2005). Unfortunately, the comment made by Ozer and Reise (1994, p. 374) is still pertinent 20 years on: “to date, applications of spectral analysis have been used solely in the analysis of mood. We look forward to seeing how the method is used to study behavioral variability in areas outside of the affective realm.” In this paper, we hope to show how this method can be used to study the mood-creativity relationship. Our participant has bipolar affective disorder, and is also a lecturer in Art—and that is probably not a coincidence (Jamison, 1993).

Method

Results

Mood Space

Table 2 presents summary statistics for each of the 16 BMIS items. As can be seen, aside from Loving, the full range of emotions is apparent. The modal profile is not lively, not happy, somewhat sad, quite tired, somewhat caring, not content, quite gloomy, quite jittery, somewhat drowsy, fluctuating in being grouchy, not peppy, not nervous, not calm, not loving, quite fed up, and not active.

Table 2. These are the 16 BMIS Adjectives/Phrases Pertaining to Our Participant's Mood, Assessed Each Morning, Together with Their Factor Structure

Item	Min	Mean (+ SD)	Mode (+ count)a	Max	Factor 1	Factor 2
Lively	1	2.48 (1.15)	2 (18)	5	.88c	.22
Happy	1	1.89 (0.90)	1 (22), 2 (19)	4	.84	−.15
Sad	1	2.61 (1.02)	3 (27)	4	−.64	−.24
Tired	1	3.15 (0.94)	3 (18), 4 (22)	5	−.51	.01
Caring	1	2.65 (1.05)	2 (17), 3 (16)	5	.78	.08
Content	1	1.63 (0.85)	1 (32)	4	.85	.07
Gloomy	1	3.09 (1.03)	3 (21), 4 (19)	5	−.80	.25
Jittery	1	3.07 (1.01)	3 (18), 4 (17)	5	.06	.82
Drowsy	1	3.00 (1.01)	3 (21)	5	−.30	.06
Grouchy	1	2.93 (1.08)	2 (16), 4 (18)	5	−.22	.82
Peppy	1	2.24 (1.15)	1 (17), 2 (18)	5	.82	.33
Nervous	1	2.78 (1.16)	2 (18)	5	.15	.78
Calm	1	2.17 (1.01)	1 (17), 2 (17)	4	−.00	−.78
Loving	1	1.73 (0.69)b	1 (21), 2 (24)	3	.40	−.17
Fed up	1	2.80 (1.22)	3 (15), 4 (15)	5	−.76	.18
Active	1	2.65 (1.22)	2 (20)	5	.82	.27

Notes.

• ^aAnd near-to, or co-mode. ^bMissing data on two occasions. ^cCriterial loadings (> 0.4) appear in bold.

The first question of interest that was investigated concerned the underlying structure of the affective space of our participant. Exploratory factor analysis incorporated the following steps: (a) an unrestricted Principal Components solution, from which the number of factors could be determined by scree test (here, 2); (b) a restricted factor analysis, with subsequent rotation (here orthogonal, given the .008 correlation between factors), using the value of 0.4 as criterion for marking those items loading on each factor, with an eye for ‘simple structure’. The results of this analysis appear in Table 2, where it can be seen that the first factor, describing 40% of the variance, is quite undifferentiated in terms of mood. Happy and Sad load with opposite weights, perhaps suggesting that these two markers for Positive Affect and Negative Affect, respectively, are indicative of a bipolar dimension of Pleasure. Unfortunately, both Lively and Tired also load with opposite weights on this same factor, indicating that the bipolar dimension of Arousal is not differentiated from that of Pleasure. Turning to the second factor, we find both Jittery and Calm, which are also indicative of a bipolar dimension of Arousal. Hence, while simple structure is achieved, it is not at all clear how to label these two factors.

A complementary approach to investigating this affective space is implemented in Figure 1, which presents the smallest-space analysis (SSA) for the 16 BMIS items (here, a two-dimensional space, although the coefficient of alienation for such a space is .056, which might not be considered to be adequate). While a circumplex is clearly seen (Russell, 1980), the ordering of affect here is problematic: Both Jittery and Calm, which are indicative of a bipolar dimension of Arousal, appear in close proximity, as do Happy and Sad, which are indicative of a bipolar dimension of Pleasure. Indeed, both positive affect and negative affect appear to be somewhat undifferentiated; furthermore, it is not at all clear how to label the two dimensions of this space.

In short, both analyses reveal that the mood space for our participant does not conform to that described by the accepted major dimensions of Arousal and Pleasure. Conceivably, this is because of her rapid mood fluctuations. The next section is devoted to analyzing these mood fluctuations.

Mood Fluctuations

Figure 2 presents the fluctuation of mood along time for our participant, for all 16 BMIS items, together with our measures of fluency, flexibility, and originality. Gloomy and Fed up are clearly positively correlated (r = .69, p < .0001), as are Lively and Caring (r = .82, p < .0001), Peppy and Active (r = .82, p < .0001), and Happy and Content (r = .75, p < .0001). Happy and Sad are clearly negatively correlated (r = −.54, p < .0001), as are Jittery and Calm (r = −.55, p < .0001). To get a full picture regarding mood fluctuation for each of the 16 BMIS items, we ran a series of spectral analyses, separately for each time series, looking at periods ranging between 2 and 27 days. Table 3 summarizes the major results. A 13.5-day periodicity was common to Jittery, Nervous, and Calm—all three clustering in the two-dimensional SSA plot seen in Figure 1. A 9-day periodicity was common to Happy, Sad, and Gloomy—all three clustering in the same SSA plot. Other common periodicities are marked in Table 3. On the basis of these, we continued with a series of bivariate spectral analyses, to assess the degree of squared coherence among select pairs of affects. Squared coherence, like r², is a measure of “goodness of fit” or “degree of synchrony” between the two time series, ranging between 0 and 1. We constrained the analysis by looking at coherence only at common peaks of spectral power (as indicated in Table 3). The pairs we chose to focus on were as follows: Jittery and Calm (13.5-day common peak), Happy and Sad (9-day common peak), Lively and Tired (6-day common peak), and Nervous and Calm (2.7-day common peak). Table 4 summarizes these results, together with data regarding phase differences between each pair of time series. Note there, in particular, that at their 9-day common peak, for Happy and Sad the squared coherence value is .64 (i.e., a coherence of .80), with a positive phase difference (2.60); the highest squared coherence for these two affects is .79 (2.16-day period), with a negative phase difference (−2.47).

Table 3. Univariate Spectral Analyses for Each of the 16 BMIS Adjectives/Phrases Pertaining to Our Participant's Mood, As Well As the Measures of Ideational Fluency, Flexibility, and Originality. The Cell Values Indicate Power for That Perioda

Period (days)	Lively	Happy	Sad	Tired	Caring	Content	Gloomy	Jittery	Drowsy	Grouchy	Peppy	Nervous	Calm	Loving	Fed up	Active	Fluency	Flexibility	Originality
27.00	0.25	0.93	0.58	1.42	0.12	3.06	1.65	0.27	1.26	1.38	4.77	0.32	2.80	1.20	7.45	8.61	36.86	19.77	5.18
18.00	2.42	1.94	3.54	0.10	0.34	2.71	0.95	3.35	0.48	3.95	1.28	0.08	0.51	0.38	0.83	1.65	3.56	1.86	4.60
13.50	1.28	0.58	3.63	0.60	1.91	3.26	0.45	5.66	0.03	2.49	4.19	6.13	6.13	0.61	1.30	4.70	0.69	2.74	0.06
10.80	0.08	1.53	1.33	1.95	1.01	2.63	3.54	0.47	0.48	0.74	1.02	4.75	0.63	0.19	9.97	2.62	8.23	6.14	0.05
9.00	0.76	3.34	6.64	2.07	0.10	2.76	7.64	3.09	1.32	1.10	0.73	0.13	0.56	0.02	2.74	0.76	2.87	4.64	3.10
7.71	0.27	2.80	1.71	1.26	0.03	0.29	2.87	1.02	4.19	1.27	0.29	3.83	0.83	0.78	0.10	1.19	1.72	0.73	2.57
6.75	7.90	2.19	9.41	2.18	2.36	1.49	3.04	1.20	0.60	3.70	5.30	1.01	1.04	0.96	2.44	1.59	6.24	1.26	1.59
6.00	10.84	5.45	0.28	5.74	4.41	3.19	3.51	1.57	4.66	1.38	4.78	1.61	0.14	0.89	4.52	7.81	9.31	3.32	7.72
5.40	0.64	0.38	0.25	4.16	0.80	0.03	0.35	1.73	5.00	1.58	0.22	6.27	0.80	0.02	0.33	1.83	4.43	0.68	0.90
4.91	2.18	0.56	0.13	2.63	5.59	1.09	1.31	0.13	3.96	0.89	2.90	3.94	1.37	0.26	2.57	3.10	1.58	0.25	0.98
4.50	3.69	0.14	2.12	0.74	5.21	0.73	1.63	0.73	2.58	2.19	1.98	0.47	0.03	1.37	0.94	0.76	6.94	2.11	1.24
4.15	3.92	2.52	0.01	2.26	1.28	3.09	2.71	0.30	1.23	0.28	4.17	4.96	0.36	1.17	4.21	1.18	3.15	1.20	0.38
3.86	0.76	3.32	0.06	0.50	1.53	0.35	0.95	8.96	0.78	2.60	0.35	2.47	1.57	1.14	0.62	3.08	5.23	0.09	0.32
3.60	0.82	0.45	3.06	0.15	2.92	0.40	3.12	3.64	0.11	0.62	1.07	0.12	1.82	0.43	3.52	3.20	13.16	4.96	1.39
3.38	3.09	2.62	0.97	1.31	2.93	1.44	0.53	0.56	0.07	4.52	0.71	1.30	3.14	0.02	1.46	0.81	11.37	1.72	1.59
3.18	4.68	1.51	0.67	5.34	3.29	0.66	1.15	1.82	4.88	2.95	5.75	2.50	3.11	0.05	0.65	3.94	0.52	4.05	0.19
3.00	0.50	0.35	0.67	0.90	1.56	1.01	1.77	0.24	0.39	0.04	0.75	0.34	1.80	3.03	1.37	0.38	5.54	2.39	0.13
2.84	0.86	0.58	1.70	1.71	0.17	0.47	1.11	0.38	1.37	3.61	0.48	0.11	0.10	0.18	3.12	0.90	14.03	3.07	2.28
2.70	4.03	0.74	1.96	5.26	1.41	0.88	0.47	2.30	0.62	2.34	6.12	11.10	6.79	0.62	1.29	1.07	18.63	4.83	3.99
2.57	4.25	1.31	4.09	0.78	3.30	0.19	1.62	0.99	2.16	0.65	6.19	0.56	0.20	0.89	4.65	3.89	1.70	1.84	0.58
2.45	0.31	2.86	2.80	2.02	0.19	3.23	7.54	1.55	4.29	4.33	2.60	7.66	0.12	2.09	8.43	1.43	2.14	4.47	0.11
2.35	1.42	0.81	0.30	1.19	0.74	0.97	0.49	2.60	1.18	0.67	1.33	0.32	2.25	1.28	2.33	5.50	1.08	1.50	1.44
2.25	1.22	0.94	2.23	1.25	0.10	0.60	0.96	1.32	1.75	5.49	1.61	2.07	3.12	0.92	0.74	2.23	4.61	3.55	0.70
2.16	3.85	0.47	3.75	0.89	1.15	0.41	0.32	6.20	3.38	3.57	3.94	3.72	5.70	0.15	0.47	0.86	0.55	1.27	1.03
2.08	0.58	0.47	0.39	0.18	1.83	0.58	0.63	1.04	2.12	1.19	1.72	2.47	1.18	0.42	2.32	1.21	4.09	1.91	0.78
2.00	0.78	0.69	0.27	0.16	2.17	0.70	0.02	0.01	1.13	0.72	0.75	0.01	0.01	0.69	4.86	0.54	0.03	0.89	0.04

Notes.

• ^a Values of comparative interest appear in bold.

Table 4. Bivariate Spectral Analyses for Select Pairs of the 16 BMIS Adjectives/Phrases Pertaining to Our Participant's Mood. The Cell Values Indicate Squared Coherence (Coh) and Phase Difference (dif) for That Perioda

Period (days)	Jittery Coh	& Calm Phase dif	Happy Coh	& Sad Phase dif	Lively Coh	& Tired Phase dif	Nervous Coh	& Calm Phase dif	Happy & Coh	Flexibility Phase dif	Sad & Coh	Flexibility Phase dif	Happy & Coh	Originality Phase dif	Sad & Coh	Originality Phase dif
27.00	0.16	−2.78	0.11	2.97	0.16	2.31	0.61	2.32	0.10	2.02	0.55	−0.61	0.72	−0.73	0.18	2.20
18.00	0.45	−2.46	0.63	3.10	0.02	2.47	0.59	2.60	0.24	2.38	0.53	−0.45	0.76	−0.51	0.35	2.48
13.50	0.70	−2.35	0.71	3.12	0.05	−0.41	0.51	2.55	0.57	2.28	0.66	−0.80	0.48	−0.56	0.29	2.33
10.80	0.41	−2.61	0.70	2.69	0.16	0.91	0.20	2.14	0.70	1.98	0.74	−0.79	0.60	−1.02	0.53	1.47
9.00	0.48	2.74	0.64	2.60	0.42	1.68	0.05	2.14	0.55	2.00	0.60	−0.39	0.83	−1.15	0.65	1.53
7.71	0.56	2.83	0.48	2.84	0.37	2.46	0.48	−2.98	0.19	2.79	0.51	0.32	0.72	−1.05	0.51	2.09
6.75	0.54	−2.81	0.43	2.82	0.42	2.29	0.55	−3.03	0.36	−2.60	0.85	1.01	0.74	−0.56	0.42	2.53
6.00	0.62	−2.65	0.32	2.85	0.44	1.90	0.34	2.36	0.26	−2.42	0.93	1.14	0.92	−0.42	0.29	2.69
5.40	0.38	−2.77	0.30	2.89	0.20	2.23	0.30	2.28	0.10	−2.25	0.65	1.07	0.93	−0.40	0.33	2.70
4.91	0.11	−2.39	0.09	3.07	0.14	−2.88	0.29	−3.13	0.38	−0.30	0.04	0.75	0.70	−0.42	0.15	−2.84
4.50	0.15	−2.73	0.06	−2.19	0.14	2.72	0.31	3.07	0.70	−0.31	0.06	−2.21	0.12	−0.94	0.45	−2.15
4.15	0.77	3.05	0.07	−2.18	0.50	2.70	0.35	2.57	0.16	−0.24	0.23	−1.95	0.06	−1.53	0.46	−2.23
3.86	0.86	2.98	0.14	−2.24	0.77	2.81	0.36	2.97	0.61	2.67	0.21	−1.56	0.16	−0.59	0.61	−3.07
3.60	0.68	3.00	0.23	−2.54	0.73	2.85	0.33	2.94	0.88	2.67	0.18	−1.10	0.41	−0.41	0.85	3.02
3.38	0.60	3.08	0.45	−3.00	0.83	3.12	0.29	2.98	0.33	2.49	0.04	−0.78	0.61	−0.29	0.84	2.89
3.18	0.61	2.86	0.67	−3.04	0.92	−3.03	0.44	−2.59	0.14	0.43	0.17	2.83	0.55	−0.41	0.71	2.60
3.00	0.41	2.51	0.31	2.90	0.84	−3.12	0.46	−2.59	0.06	−0.35	0.58	2.73	0.36	−0.12	0.61	2.15
2.84	0.46	2.92	0.46	2.13	0.83	2.69	0.69	2.65	0.21	−2.94	0.18	2.00	0.61	0.45	0.51	2.32
2.70	0.69	3.10	0.46	2.78	0.76	2.69	0.84	2.54	0.33	−3.03	0.39	0.61	0.32	0.62	0.35	2.58
2.57	0.43	3.06	0.61	−2.71	0.53	2.82	0.38	2.64	0.36	−2.18	0.75	0.60	0.01	0.59	0.14	2.23
2.45	0.26	2.00	0.73	−2.28	0.35	2.48	0.06	−2.70	0.53	−1.75	0.68	0.69	0.02	2.89	0.03	1.62
2.35	0.17	2.03	0.71	−2.12	0.64	2.16	0.21	2.99	0.20	−2.17	0.26	0.68	0.04	2.85	0.07	−2.80
2.25	0.30	3.04	0.71	−2.35	0.49	2.55	0.75	2.92	0.09	−2.92	0.10	−0.09	0.09	0.53	0.44	−2.22
2.16	0.69	2.95	0.79	−2.47	0.52	−3.10	0.81	−3.13	0.04	0.34	0.02	−2.35	0.21	0.57	0.53	−1.92
2.08	0.88	2.76	0.52	−2.53	0.57	−3.01	0.72	−2.78	0.29	0.55	0.42	−2.76	0.06	2.81	0.14	−1.43
2.00	0.69	0	0.26	0	0.39	0	0.61	0	0.19	0	0.47	0	0.49	0	0.14	0

Notes.

• ^a Values of comparative interest appear in bold.

Discussion

Using the same procedure of data collection in a previous study, it was reported there that while Pleasure and Arousal were found to be orthogonal for a male participant, whose affective space presented a partial replication of the circumplex commonly reported in the literature (Russell, 1980; Yik, Russell, & Steiger, 2011), this was not the case for a female participant (Glicksohn et al., 1995–1996). In this study, whose participant is a female presenting with bipolar affective disorder, a circumplex is clearly seen, but the ordering of affect here is problematic, with both positive affect and negative affect appearing to be somewhat undifferentiated. It would seem that rapid mood fluctuations compromise the structure of affective space. Indeed, this might well be expected for bipolar individuals, given the recent suggestion that “…what appear to be episodes of vacillation are actually episodes of simultaneously mixed emotions and that individuals with bipolar disorder experience more mixed emotions than others do” (Larsen & McGraw, 2011, p. 1108). If one adds to this the possibility of ‘ultra-ultra rapid’ mood cycling (Kramlinger & Post, 1996) in our participant (but an option not assessed in this study), then it is clear that her affective space should be markedly different from that of a more normative sample. One should, however, also note a limitation in our use of the BMIS to assess our participant's mood: To what extent these particular 16 adjectives/phrases can adequately capture the structure of the circumplex is not clear, especially given a criticism raised in the literature that terms such as Tired and Content are “insufficiently pure emotions” (as noted in the paper by Mayer, Salovey, Gomberg-Kaufman, & Blainey, 1991, p. 101). Nevertheless, given that for her Happy and Sad were found to be negatively correlated over time, and yet appear in close proximity in her affective space, it is more likely that she both fluctuates between the two and, sometimes, experiences both (a mixed emotion). How to disentangle these two options is a topic worthy of future study.

Both Happy and Sad exhibited a common 9-day periodicity with both ideational flexibility and originality, and that is an interesting finding regarding the mood-creativity correlation in bipolar affective disorder. According to Goodwin and Jamison (2007, p. 392), such “mixed mood changes” have also been reported by roughly one-fifth of a sample of eminent authors and writers, a sizeable proportion (38%) of whom had actually been treated for an affective disorder. A much faster 2.7-day periodicity was found for Tired, Peppy, Nervous, and Calm, on the one hand, and ideational fluency, ideational flexibility and originality, on the other. But let us first note two limitations of our study here, before elaborating on the potential significance of these two findings. First, we have a total of 54 values for each measure, which should be an adequate time series (Horne, Yang, & Ware, 1982, p. 179), though as Larsen (1990, p. 334) notes, “most researchers have at least 60 consecutive observations.” A caveat here is that in our data these are not truly consecutive observations, because data were collected on only six consecutive days of the week (Saturday excluded). To what extent this might seriously impact on the results is not known—in any event, we did not replace the missing data in our time series. Thus, the actual common periodicities might be altered because of this, although in practice this should probably have little bearing on the degree of rhythmic covariation that we report here. Second, our participant was given 2 minutes for each item of our version of the AU Task. A 3-minute assessment, such as employed by Martindale (2007), might have resulted in a slightly more number of uses, as might have a 2.5-minute assessment (Glicksohn, Kraemer, & Yisraeli, 1993), and a 1.5-minute assessment (Netz, Tomer, Axelrad, Argov, & Inbar, 2007) might have resulted in a slightly less number of uses. As it is, the number of uses that she reported in this study is somewhat lower than what one might expect based on the normative literature (see Table 1), but this could also be indicative of a general impairment in cognitive functioning seen in bipolar affective disorder (Martínez-Arán et al., 2004). As one anonymous reviewer of a previous version of this paper has noted, such norms are generally based on younger participants (indeed, the comparative data we report in Table 1 are taken from a previous study, employing females students ranging in age between 20 and 35 years). Thus, as this reviewer has suggested, the age of our participant could be another reason for this discrepancy in the number of uses reported.

With these limitations in mind, we now focus on the difference in common periodicity found with respect to ideational flexibility and originality (9-day periodicity together with Happy and Sad), and with respect to ideational fluency, ideational flexibility and originality (2.7-day periodicity together with Tired, Peppy, Nervous, and Calm). This is not the first time that ideational flexibility (and not ideational fluency) has been highlighted as being a key index indicating a shift in creativity (Ben-Soussan et al., 2013). Thus, its rhythmic covariation with both Happy and Sad is of particular interest. Indeed, this finding brings into question the relationship between creativity (here, both ideational flexibility and originality) and both positive and negative mood. For if being happy is viewed as a state that enhances creativity, while being sad is viewed as a state that is not inherently related to creativity (Baas et al., 2008), then the common periodicity found here with both states is problematic. There might also be a mismatch with subjective experience, in that bipolar participants prefer to associate their creativity with “high mood” (Taylor et al., 2015, p. 660). A more specific, differential hypothesis, is that being happy should increase all three of our indices—flexibility, fluency, and originality—while being sad should not influence all three (De Dreu et al., 2008, p. 742), and might not even be related to any of these. What have we found here? First, that being happy covaries with only two of these indices. Second, that being sad also covaries with these same two indices. In short, our results raise more questions than provide definitive answers.

Recall that the claim has been made in the literature that it is an increase in quantity and not necessarily quality which underlies the mood-creativity correlation in bipolar affective disorder (Weisberg, 1994). It is tantalizing to consider that this might be true for the faster 2.7-day cycle, with ideational fluency (quantity) and ideational flexibility (quality?) correlating with each other and with the ups and downs of the disorder. Conceivably, fluency is related to “elevated moods” (Peppy), while flexibility—if this indeed reflects quality—might be related to “negative mood” (Nervous), according to Vosburg (1998a,b). Indeed, fluency was found to be the single significant predictor for Peppy in the multiple-regression analysis. Yet, it is by far more interesting to consider that the fluctuations in affective tone along Pleasure, at a slower 9-day cycle, bear fruit in both ideational flexibility and originality—both of which reflect quality. Indeed, originality was found to be the simple significant predictor for Happy in the multiple-regression analysis.

There is much room for thought here and for further exploration. Indeed, the successful implementation of a within-participant design and its corresponding within-participant data-analytic procedures (Glicksohn, 2004), as presented in this paper, should encourage others to make further progress in disentangling the mood-creativity correlation in bipolar affective disorder.