Hoppa_2026-html.html

Behavioral evidence of uncertainty monitoring in

Drosophila melanogaster and its association with

mushroom body function

Jacob Hoppa

Academies of Loudoun, Academy of Science

https://orcid.org/0009-0001-2518-1464

Research Article

Keywords:

Drosophila melanogaster, Uncertainty Monitoring, Mushroom Body, Decision-Making,

Repeated-Measures ANOVA

Posted Date:

August 25th, 2026

DOI:

https://doi.org/10.21203/rs.3.rs-10806948/v1

License:





This work is licensed under a Creative Commons Attribution 4.0 International License.

Read Full License

Additional Declarations:

The authors declare no competing interests.

Behavioral evidence of uncertainty monitoring in

Drosophila

melanogaster

and its association with mushroom body function

Jacob Hoppa

Academy of Science, Academies of Loudoun, Leesburg, Virginia 20175, United States of

America
* Corresponding author

Highlights

●

Drosophila melanogaster

exhibited uncertainty-dependent behavioral responses

●

Decision and checking latencies increased as visual uncertainty increased

●

Mushroom body silencing reduced uncertainty-dependent latency modulation

●

Baseline locomotor performance remained comparable across genotypes

●

Abort and success outcomes showed descriptive uncertainty-related patterns

●

Findings support uncertainty-sensitive behavioral processing in compact neural systems

Article Info

Keywords

Drosophila melanogaster

Uncertainty Monitoring
Mushroom Body
Decision-Making
Repeated-Measures ANOVA

Abstract

Whether uncertainty-sensitive behaviors require large and complex neural systems or can emerge
from compact neural circuits remains unresolved. Metacognition, particularly uncertainty
monitoring, involves behavioral adjustment based on the reliability of available information;
however, distinguishing such responses from lower-level associative mechanisms remains
challenging in nonhuman animals. To investigate uncertainty-sensitive behavior in a genetically
tractable invertebrate model, behavioral experiments were conducted using

Drosophila

melanogaster

. Mushroom body neurons were silenced using the GAL4/UAS system

(29BD-GAL4 × UAS-Kir2.1), and flies were tested in a negative geotaxis assay in which visual
uncertainty was manipulated across Easy, Medium, and Hard contrast conditions. Behavioral
measures included decision latency, checking latency, abort behavior, and success, while separate
baseline assays assessed locomotor performance. Decision and checking latencies were analyzed
at the cohort level using mixed-design repeated-measures ANOVAs with uncertainty as the
within-cohort factor and genotype as the between-cohort factor. Increasing visual uncertainty
significantly affected both latency measures, and significant genotype × uncertainty interactions
were observed for both decision and checking latency. Control genotypes progressively increased
latency as uncertainty increased, whereas 29BD > Kir2.1 cohorts showed comparatively little
modulation across conditions. Baseline climbing latency and climbing height did not
significantly differ across genotypes. Abort and success outcomes were summarized
descriptively because sparse events, ceiling effects, and singular mixed-model estimation limited
reliable inferential analysis. Together, these findings provide behavioral evidence consistent with
uncertainty-sensitive processing in

Drosophila

and suggest that mushroom body activity

contributes to uncertainty-dependent behavioral regulation.

1. Introduction

Metacognition, particularly the ability to monitor uncertainty during decision-making,

has traditionally been associated with large and complex cortical systems. However, whether
uncertainty-sensitive behaviors require advanced vertebrate neural architecture or can emerge
from compact neural circuits remains unresolved. Understanding how organisms evaluate
uncertainty is important for clarifying the neural basis of higher-order cognition and may provide
insight into cognitive dysfunction associated with neurodegenerative diseases, psychiatric
disorders, and brain injuries (Garcia-Cordero et al., 2021).

Metacognition is commonly defined as the ability to monitor and regulate one’s own

cognitive processes (Fleur et al., 2021; Stanton et al., 2021). One measurable component of
metacognition is uncertainty monitoring, in which organisms adjust behavior based on the
perceived reliability of their decisions (Qu et al., 2023; Fleming & Lau, 2014). Uncertainty
monitoring contributes to error detection and behavioral correction, both of which are essential
for adaptive decision making in dynamic environments (Qu et al., 2023). Consequently,
uncertainty-sensitive behavior has become an important behavioral framework for investigating
metacognitive processes.

Although metacognition has been extensively documented in humans and nonhuman

primates, studying metacognitive processes in animals remains challenging (Beran, 2019). A
major difficulty involves distinguishing genuine uncertainty monitoring from low-level
associative learning mechanisms (Smith et al., 2012). Animals may appear to display
metacognitive behaviors while instead relying on reinforced stimulus-response associations. This
interpretative burden has complicated efforts to determine whether nonhuman species possess
uncertainty-sensitive cognitive processes and has limited understanding of the neural
mechanisms underlying these behaviors (Fleming & Lau, 2014).

Drosophila melanogaster

provides a valuable system for addressing these challenges

because its compact nervous system permits precise neural circuit manipulation while still
supporting complex behaviors. Despite possessing approximately 200,000 neurons compared
with roughly 100 billion neurons in humans,

Drosophila

exhibits sophisticated sensory

integration, learning, and memory capabilities (Koon & Chan, 2017; Martinez-Cervantes et al.,
2022; Rivi et al., 2023). The GAL4/UAS system further enables targeted genetic manipulation of
defined neuronal populations, allowing direct investigation of neural contributions to
cognitive-like behaviors (Ayomide Victor Atoki et al., 2024).

The mushroom body (MB) is a major neural structure involved in sensory integration,

learning, and behavioral choice in

Drosophila

(Aso et al., 2014; Cheong et al., 2020). By

silencing mushroom body neurons and examining behavioral responses to graded uncertainty,
this study directly tests whether uncertainty-sensitive behavioral regulation depends on
mushroom body activity and whether such regulation can emerge from comparatively compact
neural circuits (Barron & Mourmourakis, 2023).

To the best of our knowledge, uncertainty-sensitive behavior has not been directly

investigated in

Drosophila melanogaster

using this type of graded visual-uncertainty paradigm.

Previous work has largely focused on conditioned learning and visually guided decision-making
rather than uncertainty monitoring, latency-based hesitation behaviors, or post-choice evaluation
(Cheong et al., 2020). Addressing this gap may clarify whether metacognitive-like processes
require large neural systems or can emerge from compact circuits. Therefore, the purpose of this
study was to investigate uncertainty-sensitive decision-making behaviors in

Drosophila

melanogaster and determine whether mushroom body activity modulates these behaviors. Using
GAL4/UAS-mediated neuronal silencing and graded visual uncertainty conditions, this study
examined decision latency, checking latency, abort behavior, and locomotor performance as
behavioral indicators of uncertainty monitoring.

2. Methods and materials

2.1. Materials

Parental genotypes were obtained from Bloomington Drosophila Stock Center (BDSC).

This included the 29BD-GAL4 (Stock #: 26793) driver line and UAS-Kir2.1 (Stock #: 606319)
responder line to create the experimental flies. Flies were maintained in standard fly vials on
food composed of yeast, soy flour, yellow cornmeal, agar, corn syrup, and water. Propionic acid
was added to prevent microbial growth.

The food was prepared by measuring 6.75 g of yeast, 3.90 g of soy flour, 28.5 g of yellow

cornmeal, and 2.25 g of agar into a beaker. A mixture of 30 g of corn syrup and 390 g of water
was added, and the mixture was heated in 30-second intervals, with stirring between intervals,
until boiling. After cooling to approximately 60–70°C, 1.88 mL of propionic acid was added.
The food was poured into vials (~10 mL per vial), capped, and allowed to solidify. All other
materials used were readily available through the Academies of Science program.

2.2. Experimental design and groups

Four genotypes were tested: wild-type, 29BD-GAL4 (parental control), UAS-Kir2.1

(parental control), and 29BD > Kir2.1 (mushroom body-silenced). Each genotype consisted of
five independent cohorts of 10 flies (50 flies per genotype). Each cohort completed two trials of
the baseline locomotion assay and two trials under each uncertainty condition.

2.3. Fly preparation and crosses

Flies were sorted by sex based on abdominal shapes (round in males and pointed in

females) and sex combs (present only in males). Virgin females were identified by examining for
a clear and swollen abdomen, swollen front head, and folded wings.

To enable handling and sorting of flies by sex and virginity, flies were anesthetized using

a cold-sorting method. Cold-sorting was performed by placing a vial of approximately 10-20
flies into a bowl of ice for 2-3 minutes until immobilized. Flies were then placed on a weight
boat with ice under it, and examined using a magnifying glass and a feather. This method
minimized stress and avoided chemical anesthesia that could affect behavior.

To generate experimental flies with silenced mushroom body neurons, 5–6 virgin

29BD-GAL4 females were crossed with 4 UAS-Kir2.1 males in food vials. Crosses were
maintained at 25°C in an incubator, and parents were transferred to new vials every 3–4 days.
Progeny were collected after approximately 10 days, and flies without curly wings were selected
for experiments, while flies displaying the curly-wing balancer phenotype were discarded.

Parental control lines were generated by crossing 29BD-GAL4 with wild-type flies and

UAS-Kir2.1 with wild-type flies using the same procedure.

2.4. Baseline locomotion assay

Baseline locomotion was assessed using a negative geotaxis assay to confirm intact motor

function across genotypes. Clean vials were marked at 1 cm intervals to the top of the vial by
placing a small black dot using a Sharpie. All baseline trials were conducted under standardized
visual conditions without contrast manipulation.

To maintain a consistent background, white paper was positioned vertically

approximately 4 inches from the back and sides of the vial, while the front remained open for
video recording. This ensured uniform visual conditions while allowing clear observation of fly
movement.

Ten flies were placed into vials and allowed to acclimate for 2 minutes. Each trial began

by tapping the vial to bring flies to the bottom, followed by immediate recording for 20 seconds.
Flies were recorded using an iPhone 16 Pro on SLO-MO, HD resolution at 120 fps. Behavioral
measures included climbing latency (time to cross 3 cm), success (reaching 8 cm within 10
seconds), and climbing height (cm). Each cohort of each group underwent 2 trials with 90
seconds of rest between trials. To record latency, the stopwatch on another iPhone 16 Pro was
placed in view of recording, but out of sight for the flies, on the left side of the left piece of white
paper. The stopwatch was started before each trial began, ran through video recording, and
stopped once video recording ended.

2.5. Contrast-uncertainty assay

To test uncertainty-dependent behavior, a negative geotaxis climbing assay was

conducted under three visual contrast conditions: Easy (high contrast, low uncertainty), Medium
(moderate contrast), and Hard (low contrast, high uncertainty).

Visual uncertainty was manipulated by placing contrast sleeves around the vial at 75% of

the vial’s circumference, reducing the reliability of visual cues. Contrast sleeves were created in
Inkscape by first measuring the height and circumference of the fly vials. Height was measured
to be ~10 cm, and circumference was measured to be ~7 cm. The circumference value was then
recalculated to 75% of its original value (~5.2 cm). In Inkscape, a new file was opened, and a 2D
rectangle was made by making the length 100 mm and the width 52 mm. To create the Easy
condition, 10 x 2 mm rectangles were alternated between black (rgb(0, 0, 0)) and white
(rgb(255, 255, 255)) across the whole plane. A similar process was used to create the medium
and hard contrast; however, for the Medium condition, 10 x 2 mm rectangles were alternated

between grey (rgb(80, 80, 80)) and white (rgb(255, 255, 255)), and for the Hard condition, the
entire plane consisted of a darker grey (rgb(127, 127, 127)). Each sleeve was taped around one
vial using clear Scotch tape. Sleeves were taped and removed from the same vial for each
condition. The reason each sleeve only wrapped around 75% of the vial's circumference was to
ensure that visual uncertainty was manipulated, with a small enough gap to record fly movement.

For each trial, 10 flies were recorded using an iPhone 16 Pro on SLO-MO, HD

resolution, 120 fps for 20 seconds after being tapped on the bottom of the vial. The order of
contrast conditions was counterbalanced across trials to control for fatigue and habituation
effects. Behavioral measures included decision latency, defined as the time to cross 3 cm;
checking latency, defined as the time spent between 0–3 cm before sustained climbing; abort,
defined as the initiation followed by retreat before 3 cm; and success, defined as reaching 8 cm.
Each cohort of each group underwent 2 trials with 90 seconds of rest between individual contrast
condition trials, and 5 minutes of rest between contrast conditions. To record latency, the
stopwatch on an iPhone 16 Pro was placed in view of recording, but out of sight for the flies, on
the left side of the left piece of white paper. The stopwatch was started before each trial began,
ran through video recording, and stopped once video recording ended.

2.6. Data collection

Behavioral data were collected using recorded videos of each trial. Data collected

included genotype, trial number, contrast condition, and behavioral metrics. Behavioral
indicators such as hesitation, pausing, and movement patterns were also recorded to support
analysis of uncertainty-related behaviors.

2.7. Data analysis

All statistical analyses were conducted using Microsoft Excel and JASP 0.95.4.0

statistical software. Raw behavioral data were initially organized in long format. Because cohort
numbers 1–5 were repeated within each genotype, a Unique Cohort variable was created by
combining genotype and cohort number, producing 20 unique cohort identifiers across the four
genotypes.

For decision latency and checking latency, raw observations were aggregated in

Microsoft Excel using PivotTables. Unique Cohort was used as the row variable, uncertainty
condition (Easy, Medium, Hard) as the column variable, and latency as the value variable
summarized by the arithmetic mean. This produced one cohort-level mean for each latency
measure at each uncertainty condition. The resulting 20-row cohort-level dataset was then
imported into JASP.

Decision and checking latency were analyzed separately using mixed-design

repeated-measures ANOVAs. Uncertainty condition was specified as the within-cohort factor and
genotype as the between-cohort factor. The genotype × uncertainty interaction tested whether
uncertainty-dependent changes differed among genotypes. Mauchly’s test was used to assess
sphericity, and Greenhouse–Geisser corrections were applied when this assumption was violated.

Significant interactions were followed by simple-main-effects analyses and Holm-adjusted
conditional comparisons. Statistical significance was defined as

< .05, and effect sizes were

reported as .

Baseline climbing latency and climbing height were likewise aggregated in Excel at the

Unique Cohort level using arithmetic means and then imported into JASP. Each outcome was
analyzed separately using a one-way ANOVA with genotype as the factor. Homogeneity of
variance was assessed using Levene’s test, and residual distributions were inspected using Q–Q
plots. Because the homogeneity-of-variance assumption was violated for climbing latency,
Welch’s one-way ANOVA was used for that outcome; the standard one-way ANOVA was used
for climbing height.

Abort and success outcomes were also summarized in Excel using PivotTables with

Unique Cohort as rows and uncertainty condition as columns. Because these variables were
coded 0/1, their cohort-level means represented event proportions. A binomial generalized linear
mixed model was evaluated for abort using genotype and uncertainty as fixed effects and Unique
Cohort as a random intercept. The model produced a singular fit with the cohort-level
random-intercept variance estimated as zero, so its inferential results were not interpreted.
Success showed a strong ceiling effect, including 100% success across all cohorts in the Easy
condition. Abort and success were therefore reported descriptively using cohort-level
proportions.

Missing latency observations were treated as missing rather than as zero and were

excluded from the corresponding cohort-level mean calculations.

2.8. Safety and disposal

All experiments followed standard laboratory safety protocols. Proper personal protective

equipment was worn at all times. Flies were euthanized in a -20°C freezer and disposed of in
accordance with biohazard waste guidelines.

2.9. AI assistance

I acknowledge the use of ChatGPT (GPT-5.6 Sol, OpenAI, https://chatgpt.com/) to refine

the language and organization of the manuscript and to proofread the final draft.

cohort means per genotype, and error bars represent 95% confidence intervals across cohorts. Baseline climbing
latency and climbing height did not significantly differ among genotypes.

Decision and checking latencies showed distinct uncertainty-dependent patterns across

genotypes (Fig. 1a–b). In 29BD > Kir2.1 cohorts, mean decision latency remained nearly
unchanged across the Easy (2.846 s), Medium (2.882 s), and Hard (2.888 s) conditions. In
contrast, decision latency increased with uncertainty in each control genotype, from 2.895 to
3.693 s in Gal4-29BD cohorts, from 2.905 to 3.893 s in UAS-Kir2.1 cohorts, and from 2.816 to
3.781 s in wild-type cohorts.

Checking latency showed a similar pattern. Mean checking latency in 29BD > Kir2.1

cohorts remained relatively stable across Easy (1.676 s), Medium (1.713 s), and Hard (1.720 s)
conditions. In contrast, checking latency increased from 1.673 to 2.786 s in Gal4-29BD cohorts,
from 1.741 to 3.129 s in UAS-Kir2.1 cohorts, and from 1.553 to 2.783 s in wild-type cohorts as
uncertainty increased. These descriptive patterns are consistent with reduced
uncertainty-dependent modulation of both decision and checking latency following mushroom
body silencing.

Baseline locomotor performance was comparatively similar across genotypes (Fig. 1c–d).

Mean baseline climbing latency ranged from 6.780 s in 29BD > Kir2.1 cohorts to 7.112 s in
wild-type cohorts, while mean climbing height ranged only from 8.065 to 8.073 cm across
genotypes. The inferential analyses presented in Section 3.3 showed no significant genotype
effect for either baseline measure, supporting the conclusion that the uncertainty-dependent
latency differences were unlikely to be attributable to generalized locomotor impairment.

3.2. Repeated-measures ANOVA results

Outcome

Effect

Decision

Latency

Uncertainty

1.403, 22.455 64.152

< .001

.653

Genotype

3, 16

8.627

.001

.252

Uncertainty

Genotype

4.210, 22.455 6.732

< .001

.203

Checking

Latency

Uncertainty

2, 32

137.950

< .001

.726

Genotype

3, 16

9.818

< .001

.280

Uncertainty

Genotype

6, 32

14.620

< .001

.323

Table 1.

Repeated-measures ANOVA results for decision and checking latency. Mixed-design repeated-measures

ANOVAs were conducted with uncertainty condition as the within-cohort factor and genotype as the between-cohort
factor. Greenhouse–Geisser-corrected degrees of freedom are reported for decision latency because Mauchly’s test

indicated a violation of sphericity. Sphericity was satisfied for checking latency; therefore, uncorrected degrees of
freedom are reported. Effect sizes are reported as omega squared ( ).

Decision latency was significantly affected by uncertainty level,

(1.403, 22.455) =

64.152,

< .001, = .653, and genotype,

(3, 16) = 8.627,

= .001, = .252 (Table 1).

Importantly, the uncertainty × genotype interaction was also significant,

(4.210, 22.455) =

6.732,

< .001, = .203, indicating that the effect of increasing visual uncertainty on decision

latency differed across genotypes. Mauchly’s test indicated that the assumption of sphericity was
violated,

= .575,

= .016; therefore, Greenhouse–Geisser-corrected statistics are reported.

Follow-up simple-main-effects analyses showed that uncertainty significantly affected

decision latency in Gal4-29BD (

= .001), UAS-Kir2.1 (

< .001), and wild-type (

= .002)

cohorts, but not in mushroom body-silenced 29BD > Kir2.1 cohorts (

= .190). Holm-adjusted

conditional comparisons further showed no significant differences among Easy, Medium, and
Hard conditions in 29BD > Kir2.1 cohorts (all

= 1.000). In contrast, all three control

genotypes showed significant increases in decision latency across uncertainty conditions. Under
the Easy condition, no significant differences were observed among genotypes. Under the Hard
condition, however, 29BD > Kir2.1 cohorts exhibited significantly shorter decision latencies than
Gal4-29BD, UAS-Kir2.1, and wild-type cohorts (all

< .001), while the control genotypes

did not significantly differ from one another.

Checking latency showed a similar but stronger genotype-dependent response. There was

a significant main effect of uncertainty,

(2, 32) = 137.950,

< .001, = .726, as well as a

significant main effect of genotype,

(3, 16) = 9.818,

< .001, = .280 (Table 1). The

uncertainty × genotype interaction was also significant,

(6, 32) = 14.620,

< .001, = .323,

demonstrating that changes in checking latency across uncertainty conditions differed by
genotype. Mauchly’s test indicated that the assumption of sphericity was satisfied,

= .950,

.681; therefore, uncorrected degrees of freedom are reported.

Simple-main-effects analyses showed that uncertainty significantly affected checking

latency in Gal4-29BD, UAS-Kir2.1, and wild-type cohorts (all

< .001), but not in 29BD >

Kir2.1 cohorts (

= .166). Holm-adjusted conditional comparisons showed no significant

differences among Easy, Medium, and Hard conditions in 29BD > Kir2.1 cohorts
(all

= 1.000). In contrast, checking latency increased significantly across uncertainty

conditions in each control genotype. Under the Easy condition, no significant genotype
differences were observed. Under the Hard condition, 29BD > Kir2.1 cohorts exhibited
significantly shorter checking latencies than Gal4-29BD, UAS-Kir2.1, and wild-type cohorts (all

< .001), whereas the three control genotypes did not significantly differ from one another.

Descriptive means further illustrate these interactions. Mean decision latency in 29BD >

Kir2.1 cohorts remained nearly unchanged from the Easy (2.846 s) to Medium (2.882 s) and
Hard (2.888 s) conditions. In contrast, decision latency increased from 2.895 to 3.693 s in

Gal4-29BD cohorts, from 2.905 to 3.893 s in UAS-Kir2.1 cohorts, and from 2.816 to 3.781 s in
wild-type cohorts from Easy to Hard conditions. Checking latency showed the same pattern:
29BD > Kir2.1 cohorts remained relatively stable across Easy (1.676 s), Medium (1.713 s), and
Hard (1.720 s) conditions, whereas checking latency increased from 1.673 to 2.786 s in
Gal4-29BD cohorts, from 1.741 to 3.129 s in UAS-Kir2.1 cohorts, and from 1.553 to 2.783 s in
wild-type cohorts. Together, these results indicate that control cohorts progressively increased
both decision and checking latency as visual uncertainty increased, whereas mushroom
body-silenced cohorts showed substantially reduced uncertainty-dependent modulation of both
behavioral measures.

3.3. Baseline locomotion analyses across genotypes

(a)

Genotype

Mean (s)

29BD > Kir2.1 6.780

0.143

Gal4-29BD

6.790

0.691

UAS-Kir2.1

6.809

0.408

7.112

0.753

(b)

Genotype

Mean (cm)

29BD > Kir2.1 8.065

0.008

Gal4-29BD

8.067

0.005

UAS-Kir2.1

8.066

0.004

8.073

0.010

Table 2.

Baseline locomotion performance across genotypes. (a) Mean climbing latency under standardized visual

conditions. Values are presented as mean ± SD across five independent cohorts per genotype. Levene’s test indicated
unequal variances,

(3, 16) = 3.437,

= .042; therefore, Welch’s one-way ANOVA was used. Climbing latency did

not differ significantly across genotypes, Welch’s

(3, 7.395) = 0.267,

= .847. (b) Mean climbing height under

standardized visual conditions. Homogeneity of variance was satisfied, Levene’s

(3, 16) = 1.370,

= .288. A

one-way ANOVA showed no significant effect of genotype on climbing height,

(3, 16) = 1.306,

= .307,

= .044. Together, these analyses showed no evidence of genotype-related baseline locomotor impairment.

Baseline locomotor performance was comparable across genotypes, indicating that the

reduced uncertainty-dependent modulation observed in 29BD > Kir2.1 cohorts was unlikely to
result from generalized locomotor impairment.

3.4. Abort and success outcomes

(a)

Genotype

Easy

Medium

Hard

29BD > Kir2.1

0.000

0.020

Gal4-29BD

0.000

0.090

0.130

UAS-Kir2.1

0.000

0.080

0.150

0.000

0.090

0.110

(b)

Genotype

Easy

Medium

Hard

29BD > Kir2.1

1.000

0.960

0.930

Gal4-29BD

1.000

0.880

0.730

UAS-Kir2.1

1.000

0.860

0.720

1.000

0.860

0.700

Table 3.

Cohort-level abort and success proportions across uncertainty conditions and genotypes. (a) Mean abort

proportions and (b) mean success proportions calculated across five independent cohorts per genotype. Binary
outcomes were summarized descriptively because abort data were sparse and the binomial mixed-effects model
produced a singular fit with cohort-level random-intercept variance estimated as zero. Success data exhibited a
strong ceiling effect, including 100% success across all cohorts in the Easy condition, limiting stable inferential
modeling. Accordingly, no inferential statistical comparisons are reported for these binary outcomes.

Abort and success outcomes were summarized descriptively at the cohort level because

their distributions did not support reliable mixed-effects inference. Abort events were absent
across all genotypes in the Easy condition and remained rare in 29BD > Kir2.1 cohorts as
uncertainty increased, with mean abort proportions of 0.000, 0.000, and 0.020 under Easy,
Medium, and Hard conditions, respectively (Table 3a). In contrast, mean abort proportions under
the Hard condition reached 0.130 in Gal4-29BD, 0.150 in UAS-Kir2.1, and 0.110 in wild-type
cohorts.

Success remained at 1.000 across all genotypes in the Easy condition (Table 3b). Under

increasing uncertainty, success declined more substantially in the control genotypes, reaching
mean proportions of 0.730 in Gal4-29BD, 0.720 in UAS-Kir2.1, and 0.700 in wild-type cohorts
under the Hard condition. In comparison, 29BD > Kir2.1 cohorts maintained a mean success
proportion of 0.930 under the Hard condition. Because these binary outcomes were characterized
by sparse events and ceiling effects, these differences are presented descriptively and should not
be interpreted as statistically established genotype effects.

4. Discussion

This study investigated whether

Drosophila melanogaster

exhibits uncertainty-dependent

behavioral responses and whether mushroom body activity contributes to these behaviors.
Wild-type and parental control flies showed progressively longer decision and checking latencies
as visual uncertainty increased, whereas mushroom body-silenced flies displayed comparatively
little modulation across conditions. Together, these findings suggest that

Drosophila

behavior is

sensitive to changes in the reliability of environmental information and that mushroom body
neurons contribute to uncertainty-dependent behavioral regulation.

The observed increases in decision and checking latency under high uncertainty are

consistent with behavioral patterns associated with uncertainty monitoring in other species.
Previous studies in vertebrates and non-primate animals have shown that organisms often
hesitate, delay responding, or engage in information-seeking behaviors when faced with
uncertain conditions (Qu et al., 2023; Smith et al., 2012). In the present study, flies exposed to
reduced visual contrast displayed longer hesitation periods before sustained climbing, suggesting
increased difficulty evaluating environmental cues. These findings support the possibility that
relatively simple nervous systems are capable of generating behaviors resembling
uncertainty-sensitive decision making.

Importantly, mushroom body silencing altered these behavioral responses without

impairing baseline locomotor performance. Mushroom body-silenced flies exhibited
comparatively stable decision and checking latencies across uncertainty conditions, while control
groups showed substantial latency increases under high uncertainty. Baseline climbing latency
and climbing height remained consistent across genotypes, indicating that the observed
behavioral differences were unlikely to result from motor deficits. Since the mushroom body is
known to integrate sensory information and guide adaptive behavior (Aso et al., 2014; Cheong et
al., 2020), these findings suggest that mushroom body activity may contribute to evaluating or
responding to uncertain sensory environments.

Abort and success outcomes provided additional descriptive evidence of behavioral

changes across uncertainty conditions, although their distributions did not support reliable
mixed-effects inference. Abort events were absent in the Easy condition and became more
frequent under greater uncertainty in the control cohorts, whereas abort behavior remained
comparatively rare in 29BD > Kir2.1 cohorts. Similarly, success remained at ceiling under the
Easy condition but declined under greater uncertainty in the control genotypes, while remaining
comparatively high in mushroom body-silenced cohorts. Because abort events were sparse and
success outcomes exhibited strong ceiling effects, these patterns should be interpreted
descriptively rather than as statistically established genotype effects. The greater sensitivity and
statistical stability of the latency measures suggest that continuous behavioral measures may be
particularly useful for examining uncertainty-dependent modulation in this paradigm.

One major challenge in animal metacognition research is distinguishing uncertainty

monitoring from associative learning explanations (Smith et al., 2012). In many vertebrate
paradigms, animals may learn fixed stimulus-response associations rather than dynamically
evaluate uncertainty during decision making. Several aspects of the present design help mitigate,
although not fully eliminate, this interpretative concern. First, uncertainty was manipulated
gradually through visual contrast conditions rather than through explicit reward contingencies,
reducing the likelihood that flies relied solely on reinforced stimulus-response rules. Second,
behavioral measures included continuous latency-based variables, including decision latency and
checking latency, rather than binary choice outcomes alone. These latency measures allowed
detection of graded hesitation behaviors that scaled with uncertainty level. Third, mushroom
body silencing reduced uncertainty-dependent modulation of both decision and checking latency
while baseline locomotion remained intact, suggesting that the observed effects were not simply
driven by generalized motor deficits. Although associative mechanisms cannot be entirely
excluded, the present findings support the interpretation that flies dynamically adjusted their
behavior in response to uncertainty-related sensory ambiguity.

Several limitations should be considered when interpreting these findings. First, the

observed behavioral changes are consistent with uncertainty-sensitive processing but do not by
themselves demonstrate metacognition. Alternative mechanisms, including changes in visual
processing, motivation, sensory integration, or action selection, could also contribute to the
observed latency patterns. Second, the present study used a relatively small number of
independent cohorts, with five cohorts per genotype, which may limit the precision and
generalizability of the estimated effects. Third, abort and success outcomes were characterized
by sparse events and ceiling effects, preventing reliable inferential analysis of these measures.
Finally, the present paradigm manipulates uncertainty through visual contrast alone; future
studies using additional sensory modalities, alternative uncertainty manipulations, and
complementary neural manipulations would help determine whether the observed effects
generalize beyond the current assay.

Overall, this study provides evidence consistent with uncertainty-sensitive behavioral

modulation in

Drosophila melanogaster

and suggests that mushroom body neurons contribute to

these responses. Although the observed latency patterns do not by themselves establish
metacognition, they indicate that flies adjust their behavior in response to changes in sensory
uncertainty and that this adjustment is reduced following mushroom body silencing. These
findings support the use of

Drosophila

as a tractable model for investigating the neural

mechanisms underlying uncertainty-related behavioral regulation and adaptive decision-making.
More broadly, this model may help clarify how compact neural circuits integrate uncertain
sensory information and how disruption of these processes may alter flexible behavior.

5. Conclusion

This study provides evidence of uncertainty-dependent behavioral modulation in

Drosophila melanogaster

and suggests that mushroom body neurons contribute to these

responses. As visual uncertainty increased, wild-type and parental control flies showed longer
decision and checking latencies, whereas mushroom body-silenced flies displayed reduced
behavioral adjustment across conditions. Importantly, baseline locomotion remained intact across
genotypes, suggesting that these effects were unlikely to be explained by generalized motor
impairment. Together, these findings suggest that relatively simple neural systems may support
uncertainty-sensitive processing and establish

Drosophila

as a promising model for investigating

the neural mechanisms underlying uncertainty monitoring and higher-order cognition.

Statement of Novelty

To the best of our knowledge, this study is among the first to directly investigate

uncertainty-sensitive behavioral modulation and its association with mushroom body function in

Drosophila melanogaster

. The findings show that increasing visual uncertainty alters behavioral

latency responses and that mushroom body silencing reduces uncertainty-dependent modulation
of both decision and checking behavior. These results provide novel evidence consistent with
metacognitive-like behavioral processes in an organism with a compact nervous system, while
not excluding alternative associative or sensory explanations.

Credit Authorship Contribution Statement

Jacob Hoppa:

Funding acquisition, Conceptualization, Methodology, Resources, Data

curation, Formal analysis, Writing – original draft, Writing – review & editing.

Declaration of Competing Interests

The author declares that there were no competing interests that might have affected the

work in this paper.

Acknowledgments

This study was supported by the Academies of Science research program of the

Academies of Loudoun in the United States.

Grant Information

The author declares that no grants were involved in supporting this work.

Supporting information

All supplementary data used in this article can be found in Zenodo

(

https://doi.org/10.5281/zenodo.21962583

References

Aso, Y., Sitaraman, D., Ichinose, T., Kaun, K. R., Vogt, K., Belliart-Guérin, G., Plaçais, P.-Y.,

Robie, A. A., Yamagata, N., Schnaitmann, C., Rowell, W. J., Johnston, R. M., Ngo, T.-T.

B., Chen, N., Korff, W., Nitabach, M. N., Heberlein, U., Preat, T., Branson, K. M., &

Tanimoto, H. (2014). Mushroom body output neurons encode valence and guide

memory-based action selection in Drosophila.

ELife

https://doi.org/10.7554/elife.04580

Ayomide Victor Atoki, Aja, P. M., Tijjani Salihu Shinkafi, Ondari, E. N., Adeniyi, A. I.,

Ilemobayo Victor Fasogbon, Dangana, R. S., Shehu, U. U., & Akinpelumi

Akin-Adewumi. (2024). Exploring the versatility of

Drosophila melanogaster

as a model

organism in biomedical research: a comprehensive review.

Fly

(1).

https://doi.org/10.1080/19336934.2024.2420453

Barron, A. B., & Mourmourakis, F. (2023). The Relationship between Cognition and Brain Size

or Neuron Number.

Brain Behavior and Evolution

(2), 109–122.

https://doi.org/10.1159/000532013

Beran, M. (2019). Animal metacognition: A decade of progress, problems, and the development

of new prospects.

Animal Behavior and Cognition

(4), 223–229.

https://doi.org/10.26451/abc.06.04.01.2019

Cheong, H. S., Siwanowicz, I., & Card, G. M. (2020). Multi-regional circuits underlying visually

guided decision-making in Drosophila.

Current Opinion in Neurobiology

, 77–87.

https://doi.org/10.1016/j.conb.2020.10.010

Fleming, S. M., & Lau, H. C. (2014). How to measure metacognition.

Frontiers in Human

Neuroscience

(1). https://doi.org/10.3389/fnhum.2014.00443

Fleur, D. S., Bredeweg, B., & van den Bos, W. (2021). Metacognition: ideas and insights from

neuro- and educational sciences.

Npj Science of Learning

(1), 1–11.

https://doi.org/10.1038/s41539-021-00089-5

Garcia-Cordero, I., Migeot, J., Fittipaldi, S., Aquino, A., Campo, C. G., García, A., & Ibáñez, A.

(2021). Metacognition of emotion recognition across neurodegenerative diseases.

Cortex

137

, 93–107. https://doi.org/10.1016/j.cortex.2020.12.023

Koon, A. C., & Chan, H. Y. E. (2017). Drosophila melanogaster As a Model Organism to Study

RNA Toxicity of Repeat Expansion-Associated Neurodegenerative and Neuromuscular

Diseases.

Frontiers in Cellular Neuroscience

https://doi.org/10.3389/fncel.2017.00070

Martinez-Cervantes, J., Shah, P., Phan, A., & Cervantes-Sandoval, I. (2022). Higher-order

unimodal olfactory sensory preconditioning in Drosophila.

ELife

https://doi.org/10.7554/elife.79107

Rivi, V., Benatti, C., Rigillo, G., & Maria, J. (2023). Invertebrates as models of learning and

memory: investigating neural and molecular mechanisms.

The Journal of Experimental

Biology

226

(3). https://doi.org/10.1242/jeb.244844

Qu, Z., Shi, L., Chun, B., Yin, J., & Sze Chai Kwok. (2023). Uncertainty monitoring and

information seeking in non-primate animals: Meta-analysis and systematic review.

Frontiers in Ethology

. https://doi.org/10.3389/fetho.2023.1246370

Smith, J. D., Couchman, J. J., & Beran, M. J. (2012). The highs and lows of theoretical

interpretation in animal-metacognition research.

Philosophical Transactions of the Royal

Society B: Biological Sciences

367

(1594), 1297–1309.

https://doi.org/10.1098/rstb.2011.0366